American University in Cairo

AUC Knowledge Fountain

Theses and Dissertations Student Research

Spring 2-14-2021

Latency Optimization in Smart Meter Networks

Amr Kassab
[email protected]

Follow this and additional works at: https://fount.aucegypt.edu/etds

Part of the Systems and Communications Commons

Recommended Citation

APA Citation
Kassab, A. (2021).Latency Optimization in Smart Meter Networks [Master's Thesis, the American
University in Cairo]. AUC Knowledge Fountain.
https://fount.aucegypt.edu/etds/1611
MLA Citation
Kassab, Amr. Latency Optimization in Smart Meter Networks. 2021. American University in Cairo, Master's
Thesis. AUC Knowledge Fountain.
https://fount.aucegypt.edu/etds/1611

This Master's Thesis is brought to you for free and open access by the Student Research at AUC Knowledge
Fountain. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of AUC
Knowledge Fountain. For more information, please contact [email protected].
 A MERICAN U NIVERSITY IN C AIRO

M ASTER OF S CIENCE T HESIS

Latency Optimization in Smart Meter Networks

Author: Supervisor:
Amr Salah Kassab Dr. Karim Seddik
Dr. Ayman Elezabi

A thesis submitted in partial fulfillment of the requirements

for the degree of Master of Science

in Electronics and Communications Engineering

February 19, 2021

 iii

Declaration of Authorship
I, Amr Salah Kassab, declare that this thesis titled, “Latency Optimization
in Smart Meter Networks” and the work presented in it are my own. I
confirm that:

• This work was done wholly or mainly while in candidature for a re-
search degree at this University.

• Where any part of this thesis has previously been submitted for a de-
gree or any other qualification at this University or any other institu-
tion, this has been clearly stated.

• Where I have consulted the published work of others, this is always

clearly attributed.

• Where I have quoted from the work of others, the source is always
given. With the exception of such quotations, this thesis is entirely
my own work.

• I have acknowledged all main sources of help.

• Where the thesis is based on work done by myself jointly with others,
I have made clear exactly what was done by others and what I have
contributed myself.

Signed:

Date:
 v

AMERICAN UNIVERSITY IN CAIRO

Abstract
School of Sciences and Engineering
Electronics and Communications Engineering Department

Master of Science

Latency Optimization in Smart Meter Networks

by Amr Salah Kassab

In this thesis, we consider the problem of smart meter networks with data
collection to a central point within acceptable delay and least consumed
energy. In smart metering applications, transferring and collecting data
within delay constraints is crucial. IoT devices are usually resource-constrained
and need reliable and energy-efficient routing protocol. Furthermore, me-
ters deployed in lossy networks often lead to packet loss and congestion.
In smart grid communication, low latency and low energy consumption
are usually the main system targets. Considering these constraints, we pro-
pose an enhancement in RPL to ensure link reliability and low latency. The
proposed new additive composite metric is Delay-Aware RPL (DA-RPL).
Moreover, we propose a repeaters’ placement algorithm to meet the latency
requirements. The performance of a realistic RF network is simulated and
evaluated. On top of the routing solution, new asynchronous ordered trans-
mission algorithms of UDP data packets are proposed to further enhance
the overall network latency performance and mitigate the whole system
congestion and interference.
Experimental results show that the performance of DA-RPL is promis-
ing in terms of end-to-end delay and energy consumption. Furthermore,
the ordered asynchronous transmission of data packets resulted in signifi-
cant latency reduction using just a single routing metric.
 vii

Acknowledgements
I wish to express my great love and special regards to my family who
kept encouraging me till I reached the goal of this project.
I would like to express my deepest gratitude to my thesis supervisors,
Dr. Karim Seddik and Dr. Ayman Elezabi. They guided me to be profes-
sional and gave me persistent support along my master’s study. The tech-
nical contribution and funding of El-Sewedy Electrometer Group (EMG) is
truly appreciated.
 ix

Contents

Declaration of Authorship iii

Abstract v

Acknowledgements vii

List of Figures xiii

1 Introduction 1
1.1 Smart meters network . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Related work . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Organization of the thesis . . . . . . . . . . . . . . . . . . . . 5
1.3 List of contributions of this thesis . . . . . . . . . . . . . . . . 6

2 Smart Meter Communication 7

2.1 Review of PRIME PLC specifications . . . . . . . . . . . . . . 7
2.1.1 Physical layer . . . . . . . . . . . . . . . . . . . . . . . 7
Factors affecting physical performance . . . . . . . . 8
2.1.2 MAC layer . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Limitations of PLC . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 The challenges of PLC with AMI . . . . . . . . . . . . 9
2.3 Wireless metering technologies and standards . . . . . . . . 9
2.3.1 Wireless Meter Bus . . . . . . . . . . . . . . . . . . . . 9
2.3.2 ZIGBEE . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3.3 LoRa . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3.4 Wi-SUN . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.5 Wi-MAX . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3.6 LTE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
x

2.3.7 6LowPAN . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Hybrid RF and PLC network . . . . . . . . . . . . . . . . . . 15
2.4.1 Previously Proposed solutions . . . . . . . . . . . . . 17

3 Realistic wireless smart-meter network optimization 19

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2 Routing requirements for the AMI . . . . . . . . . . . . . . . 19
3.3 Routing protocol for low power and lossy networks (RPL) . 21
3.3.1 Overview of RPL . . . . . . . . . . . . . . . . . . . . . 21
3.3.2 CONTIKI RPL . . . . . . . . . . . . . . . . . . . . . . . 22
3.3.3 RPL with RF . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3.4 RPL with PLC . . . . . . . . . . . . . . . . . . . . . . . 25
3.3.5 RPL simulation . . . . . . . . . . . . . . . . . . . . . . 26
A. Network Simulator 3 - NS3 . . . . . . . . . . . . . 27
B. COOJA - Contiki network simulator . . . . . . . . 27
C. Whitefield simulation environment . . . . . . . . . 29
3.3.6 System model and evaluation results . . . . . . . . . 29
I. System model . . . . . . . . . . . . . . . . . . . . . . 29
II. Performance evaluation . . . . . . . . . . . . . . . . 32

4 Asynchronous Transmission of Data Packets 39

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 Proposed algorithms . . . . . . . . . . . . . . . . . . . . . . . 40
4.2.1 The rationale for proposing new algorithms . . . . . 40
4.2.2 Algorithm 1 . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2.3 Algorithm 2 . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Comparison between RF and PLC . . . . . . . . . . . . . . . 42
4.3.1 Wireless RF network . . . . . . . . . . . . . . . . . . . 42
4.3.2 PRIME PLC network . . . . . . . . . . . . . . . . . . . 45
4.3.3 RF against PLC performance . . . . . . . . . . . . . . 48

5 Conclusions and Future Work 49

5.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
 xi

5.2 Future directions . . . . . . . . . . . . . . . . . . . . . . . . . 50

 xiii

List of Figures

3.1 El-Minya PRIME network. . . . . . . . . . . . . . . . . . . . . 30

3.2 El-Minya network topology in COOJA simulator. . . . . . . 31
3.3 Nodes end-to-end delay. . . . . . . . . . . . . . . . . . . . . . 33
3.4 El-Minya network topology after the insertion of extra nodes. 35
3.5 Nodes end-to-end delay. . . . . . . . . . . . . . . . . . . . . . 35
3.6 Network average end-to-end delay. . . . . . . . . . . . . . . . 36
3.7 Nodes average power consumption. . . . . . . . . . . . . . . 37
3.8 Network average power consumption for four different rout-
ing metrics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.1 End-to-end delay comparison for RF network between syn-

chronous, random asynchronous, and both algorithms . . . 43
4.2 E2E delay performance comparison for some RF network nodes 44
4.3 End-to-end delay comparison for PLC network between syn-
chronous, random asynchronous, and both algorithms . . . 46
4.4 E2E delay performance comparison for some PLC network
nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
 xv

List of Abbreviations

AMI Advanced Metering Infrastructure

AODV Ad hoc On demand Distance Vector
ARQ Automatic Repeat Request
CFN Common Forwarding Node
CFP Contention Free Period
CSMA Carrier Sense Multiple Access
DAO Destination Advertisement Object
DCU Data Concentrator Unit
DIO DODAG Information Object
DIS DODAG Information Solicitation
DODAG Destination Oriented Directed Acyclic Graph
DSSS Direct Sequence Spread Spectrum
ETX Expected Transmission Count
LLN Low power and Lossy Network
LTE Long Term Evolution
LoRa Long Range
MIMO Multiple Input Multiple Output
MRHOF Minimum Rank with Hysteresis Objective Function
MRM Multi-path Ray tracer Medium
OF Objective Function
OFDM Orthogonal Frequency Division Multiplexing
OLSR Optimized Link State Routing
PER Packet Error Rate
PRIME PoweRline Intelligent Metering Evolution
PSD Power Spectral Density
RDC Radio Duty Cycling
xvi

RPL Routing Protocol for Low power and lossy networks (LLNs)
RSSI Received Signal Strength Indicator
6LowPAN IPv6 over Low-power Private Area Network
SM Smart Grid
UDGM Unit Desk Graph Medium
UDP User Datagram Protocol
Wi-MAX Worldwide Interoperability for Microwave Access
Wi-SUN Wireless Smart Utility Network
WM-BUS Wireless Meter BUS
WSN Wireless Sensor Network
 1

Chapter 1

Introduction

1.1 Smart meters network

Recently, there has been an increased interest in the field of the internet of
things (IoT). One of the use cases of IoT is smart grid communications, and
one of the essential elements of smart grid communication is the Advanced
Metering Infrastructure (AMI). AMI consists of smart electric meters used
to mainly monitor power consumption status and enable the collection of
real-time measurements. The monitoring process necessitates sending use-
ful information to a sink or aggregation node within an acceptable delay.
In recent years, most smart metering networks have used more than one
communication technology to achieve better system performance. Each of
those has its own specifications and characteristics. While its low deploy-
ment cost characterizes the power line communication (PLC) technology,
it is sensitive to interference and impulse noise that degrades the system
reliability [1]. While the wireless radio frequency (RF) network may also
suffer from noise and interference, it offers higher data rate transmissions
but with lower reach-ability. Therefore, a hybrid RF and PLC network is
recommended as a general solution to achieve more system reliability and
extended coverage distances [2]. It is important for large realistic networks
to dynamically enhance the network performance to meet the updated sys-
tem constraints. Low Power and Lossy Networks (LLNs) [2], like wired
PLC and wireless systems, aim to minimize the end-to-end delay and en-
ergy consumption. For a single-technology smart metering network, the
2 Chapter 1. Introduction

predefined requirements may be achieved by adopting and adapting cur-

rent system architecture. Hence, network performance should be first simu-
lated and evaluated sufficiently before the implementation and deployment
process. This may also give a good insight for future integration with other
wireless technologies. Routing protocol for low-power and lossy networks
(RPL) seems to be the most common choice to enhance the performance of
LLNs [3]. In [4], a new RPL objective function is introduced that simulates
the battery consumption of nodes and hence aims to increase the network
lifetime. The remaining energy of nodes is used in a large simulated net-
work and evaluated in [5]. In [6], a new RPL additive metric is proposed
that aims to trade-off between node delay, link reliability, and the consumed
energy. It is proved that ensuring the reliability of paths will decrease the
congestion and thus latency to reach the sink. In [7], a new composite PRL
metric is proposed that takes into account delay, hop count, link quality,
and expected transmission count (ETX). Furthermore, [8] proposed a new
RPL metric that is aware of congestion and can select the paths according to
nodes’ dynamic traffic. This approach contributes to congestion mitigation
as well as latency reduction.

In this thesis, we propose and evaluate the performance of a new delay-

and energy-aware RPL protocol on a real AMI network (which is currently
implemented using the PLC PRIME standard). The network topology is
simulated as a wireless network to give a vision for a possible future combi-
nation of wired and wireless links. We use the COOJA simulator to simulate
our wireless network, which is used with Multi-path Ray Tracer Medium
(MRM) radio environment. We experiment with different objective func-
tions in order to limit the latency under a prespecified threshold. Our mod-
ified objective function also results in reduced energy consumption, as we
will see later in the coming thesis chapters. Our preliminary results give
insight into future enhancements, adoption, and combining RF and PLC
links into a single hybrid routing metric. Moreover, we consider the place-
ment of repeater nodes in the network to meet the latency constraint. The
1.1. Smart meters network 3

repeater nodes will be used to offload the traffic at the congested network
nodes, which is the main cause of violating the latency constraints. Finally,
we proposed a new asynchronous transmission algorithm for nodes’ data
packets that contributes to the latency reduction on top of the routing solu-
tion.

1.1.1 Related work

Low Power and Lossy Networks (LLNs) is a group of interconnected em-

bedded devices with limited power, processing, and storage. Such limited
network resources make the selection of an appropriate routing protocol a
challenging task. As the hybrid network architecture exhibits the charac-
teristics of LLNs, some constraints and unique routing challenges should
be taken into account. The varied applications mean different character-
istics and conflicting requirements such as latency, reliability, and traffic
overhead. Furthermore, the dominant communication pattern in LLNs is
multi-point to point (MP2P) as well as alternating to other patterns like P2P
and P2MP. Also, the operation of LLNs exhibits deployments with different
densities, i.e., scalability issues, and hence the routing should handle rele-
vant, diverse cases. The power constraints in LLNs force the routing to en-
sure low energy consumption that sends useful updates to refresh routes. A
routing protocol that can achieve the LLNs requirements is the RPL. There
are several routing protocols used for wireless mesh networks. The rout-
ing in power line communication (PLC) may inherit some of those proto-
cols to optimize path selection according to its unique specification. For
instance, layer 3 of the PRIME PLC standard requires proactive routing.
Another study in [9] presents a comparative analysis of different routing
protocols in WSNs. Ad hoc On-Demand Distance Vector Routing Proto-
col (AODV) exhibits the highest delay and average energy consumption
among all. Optimized Link State Routing Protocol (OLSR) is character-
ized by its low power consumption, packet loss, and average delay. The
network characteristics basically control the choice of routing algorithm.
For power saving, the Power-Aware Multi-Access (PAMAS) protocol is a
4 Chapter 1. Introduction

good choice. However, for fault-tolerance purposes, the Temporally Or-

dered Routing Algorithm (TORA) can be the optimum path selection algo-
rithm. Few papers compared RPL to other related protocols. The results of
[10], [11], showed that RPL outperforms its counterpart; The Lightweight
On-demand Ad hoc Distance-vector Routing Protocol – Next Generation
(LOADng) in terms of overhead, delays, memory, and route formation. The
study in [12] also proved that RPL exhibits lower delay and higher packet
delivery ratio (PDR) than AODV.
Since our smart metering network is considered a type of LLN, we used
RPL protocol to improve its performance. Furthermore, RPL does not im-
pose the use of a specific default objective function and opens the door to
select it according to the defined application and system constraints [13].
The recent physical alliance release, called Wi-SUN with its Field Area Net-
work (FAN) profile, contains the RPL in its network layer and is strongly
recommended as a solution for the hybrid network performance enhance-
ment. The authors in [14] suggested using multiple RPL instances inspired
by existing RPL mechanisms to choose the best interface for routing per
node. The same paper suggests Parent Oriented Design (PO) that aims to
merge the two technologies into a single hybrid metric. Another study in
[15] introduced the concept of MAC layer extension for both RF and PLC
technologies. It basically targets applying some modifications to the RF
frame structure without changing MAC components. Carrier Sense Multi-
ple Access (CSMA) is applied conventionally in the contention-free period
(CFP), which represents the interval of all guaranteed time slots of all net-
work nodes. At the same time, RF CSMA is to be applied in the SCP to
replace scheduled channel polling.
1.2. Organization of the thesis 5

1.2 Organization of the thesis

This thesis presents some work and contributions in network optimiza-

tion and performance enhancement in terms of end-to-end delay and en-
ergy consumption. A new proposed composite RPL metric is used to im-
prove the system communication in the network layer. Also, a novel asyn-
chronous transmission scheme is proposed for PLC and wireless AMI net-
works to reduce the meters’ latency.
The thesis is organized as follows:

• Chapter (2) presents a literature review of the Power Line Communi-

cation (PLC) systems and discusses different wireless metering tech-
nologies used to enhance the performance of Low Power and Lossy
Networks (LLNs). Also, it presents the proposed hybrid RF and PLC
network model and provides the suggested solutions to optimize the
hybrid model performance by exploiting the advantages of both RF
and PLC technologies.

• Chapter (3) discusses the main requirements and constraints of smart

metering applications. Moreover, a new enhancement to standard
RPL termed Delay-Aware (DA-RPL) is presented and evaluated on
a realistic wireless network to ensure low nodes’ latency as well as
link reliability. We also propose repeaters’ placement algorithm to
provide sufficient congestion mitigation and meet a predefined delay
threshold on a node basis. We show that the performance of DA-RPL
is promising in terms of end-to-end delay and energy consumption.

• Chapter (4) proposes a novel asynchronous transmission scheme for

PLC and wireless AMI networks. By controlling the transmission
times of data packets, we show that we can achieve significant re-
ductions in the end-to-end delay and meet latency requirements. The
idea of the transmission is based on avoiding anticipated congested
routes by referring to the synchronous constructed Destination Ori-
ented Directed Acyclic Graph (DODAG).
6 Chapter 1. Introduction

• Chapter (5) provides the conclusions of this thesis and proposes some
directions for future research.

1.3 List of contributions of this thesis

• We introduce a new RPL additive routing metric called Delay Aware

DA-RPL in order to achieve the target latency constraints and save
consumed energy.

• We propose to insert some repeater nodes in the simulation environ-

ment to enhance the latency performance.

• We compare the performance of DA-RPL with three different objec-

tive functions. Interestingly, we show that DA-RPL demonstrated the
best latency and energy performance and succeeded in meeting the
predefined system requirements.

• We propose to use the asynchronous transmission scheme of the data

UDP packets to mitigate the expected congestion, decrease packet loss
rate, and improve the end-to-end delay for each node or smart meter
in the network. We prove that this transmission scheme is sufficient
for significant delay reduction and traffic overhead alleviation.
 7

Chapter 2

Smart Meter Communication

This chapter is divided into three sections. In the first section, we discuss
the general PRIME Power Line Communication PLC specifications in de-
tail and present the PLC restrictions. It also discusses common wireless
metering technologies used to enhance the performance of Low Power and
Lossy Networks (LLNs). A proposed hybrid RF and PLC network is pro-
vided along with its suggested methods to optimize the hybrid system per-
formance by exploiting the advantages of both RF and PLC technologies.

2.1 Review of PRIME PLC specifications

2.1.1 Physical layer

PoweRline Intelligent Metering Evolution (PRIME) is mainly the most com-

mon type of narrowband power line communication and is considered one
of the best mature technologies in automated smart meter reading. Its phys-
ical layer is based on Orthogonal Frequency Division Multiplexing (OFDM)
with a carrier scheme that depends on differential phase-shift keying. The
PRIME sampling frequency is located at 250 KHz with carrier frequencies
ranging from 42 to 89 KHz. PRIME provides high data rates that range
from 40 Kbps up to 1 Mbps. The PRIME service nodes tend to form a tree
topology rooted at the sink, called the base node. It supports bi-directional
communication and has high immunity to noise and interference thanks to
its unique modulation technique [16].
8 Chapter 2. Smart Meter Communication

Factors affecting physical performance

Many factors affect the PRIME performance in which the impedance and
line noise are the most harmful ones. A careful design should consider the
impact of the impulse noise and interference as well as the network size to
ensure a reliable network. The PRIME specifies an inter-leaver to mitigate
burst noise and enhance system performance [17].

2.1.2 MAC layer

PRIME network topology construction begins with the root base node (BN),
which broadcasts periodic beacons to all service nodes. Once received,
nodes registration and route selection processes occur simultaneously. How-
ever, PRIME does not define any maintenance criteria to the formed tree
topology to keep it up-to-date. Instead, it tends to send periodic control
’keep alive’ messages to monitor the registered nodes. This technology ba-
sically uses Automatic Repeat Request (ARQ) as a recovery mechanism to
manage and control the number of re-transmissions for unacknowledged
packets and uses a predefined timeout to cancel the registration of nodes
from topology rooted to the base node. In addition, PRIME supports multi-
hop proactive routing.

2.2 Limitations of PLC

power lines do not necessarily offer a secure media, and it is sensible to

disturbance. Its low transmission speed characterizes PLC, and this is a
critical point in smart metering applications. With the presence of numer-
ous elements on a power line network, data attenuation is likely to be an
issue. High costs of residential appliances: the cost of a power line network
modem is not always competitive with the cost of a standard modem used
to connect to a phone line network. The greater amount of electrical noise
on the line limits practical transmission speed.
2.3. Wireless metering technologies and standards 9

2.2.1 The challenges of PLC with AMI

The Advanced Metering Infrastructure (AMI) is an important part of the

Smart Grid Communication (SMG), where the Smart Meters (SM) are con-
nected and rooted to a Data Concentrator Unit (DCU) that collects user en-
ergy billing information through dynamic bi-directional communication.
In general, the DCU functions as a gateway between the Wide Area Net-
work (WAN) and the smart meters. Since PLC is sensible to noise and inter-
ference, it does not achieve the required specification of a high-percentage
coverage area and reading speed. Furthermore, the PLC-based mainly on
OFDM and use narrowband (NB) frequency is more vulnerable to signal
reduction due to impulsive noise, background noise, and Narrow-Band In-
terference (NBI). On the other hand, PLC technology has a full hold on the
utilities, whereas its RF counterpart only uses public frequency bands for
its operation.

2.3 Wireless metering technologies and standards

2.3.1 Wireless Meter Bus

Wireless Meter Bus (WM-BUS) is a European standard used to manage the

communication between different smart metering applications. It is gener-
ally characterized by its robust, long-range as well as efficient power de-
sign. It operates in the license-free ISM bands 169 MHz, 433 MHz, and 868
MHz. The latter band reflects the main advantage of wireless meter bus.
It provides a good radio frequency range with suitable antenna size and
mitigates such interference with higher communication distances [18].

WM-Bus has its unique star network topology and provides six transfer
modes representing particular applications [19]. The basic WM-Bus stack
includes physical, data link, and application layers, which can be extended
for advanced security purposes. The 868 MHz band offers high penetration
10 Chapter 2. Smart Meter Communication

through concrete walls, which alleviates the suppression of transmitted sig-

nals. The WM-Bus solution provides reliable hardware and firmware for
smart grid applications. High-performance implementation’s core compo-
nents are low power microcontroller (MCU), high-performance transceiver,
a modular software stack, and suitable deployment tools to evaluate and
deploy the solution. Wireless meter bus technology is originated from the
metering bus (M-Bus). Although it offers implementation from the physi-
cal layer to the application layer and left the network layer without a de-
fined specification, it provides an important environment to use it in the
industry. Therefore, many companies tend to design their own network
specification that suits their predefined requirements by manipulating the
embedded stack. Usually, smart electric meters are placed near households,
and there is always a need for mutual collaboration between distributors to
meet coverage and power quality needed. It is beneficial for system perfor-
mance to adapt a wireless meter bus that can also operate as a concentrator
to collect data efficiently. Considering all deployment and implementation
scenarios, no wireless technology, even wireless meter bus, will win in all
scenarios and provide good performance in all aspects.
However, using a wireless meter bus enables the transmitted signal to
seep through concrete walls with consuming lower energy. This feature is
thanks to the ability of WM-Bus to operate in the 868 MHz band. Further-
more, this band offers a good coverage area of transmission, avoids GSM
interference, enables reasonable antenna diameter, and provides a sufficient
radio frequency range.

2.3.2 ZIGBEE

It is a standard that is commonly used in wireless sensor networks and

is based on IEEE 802.15.4 specification. It is mainly specified by its low
power, low data rate, low cost, and easy deployment. ZIGBEE operates in
three ISM bands; 868 MHz, 915 MHZ, and 2.4 GHz, with data rates of 20
Kbps, 40 kbps, and 250 Kbps, respectively [20]. It basically supports three
2.3. Wireless metering technologies and standards 11

network topologies; star, mesh, and tree and provides secure and inter-
operable communication between end devices. ZIGBEE stack comprises
physical and MAC layers defined by IEEE 802.15.4 and network and appli-
cation layers specified by the ZIGBEE alliance [21]. It uses Direct Sequence
Spread Spectrum (DSSS) transmission technique and allows multiple chan-
nel access with CSMA/CA. Thanks to its capability of supporting a large
number of nodes, ZIGBEE is used in many real-time monitoring and con-
trol systems.
ZigBee is known to have several significant advantages, including but
not limited to the delivery of customers’ use of energy, gas, and water in an
automated, monitored, and controlled manner. It can also be used to moni-
tor patient status and many customer devices like computers and their fea-
tures. It enables interoperability between different wireless technologies.
Furthermore, it offers up-to-date pricing improvements, as well as prepaid
services. However, ZigBee is not the best candidate to support communica-
tion through concrete walls as it is commonly used in the 2.4 GHz band.
Moreover, it still provides poor network synchronization that consumes
and wastes nodes’ power.

2.3.3 LoRa

LoRa is a long-range wireless protocol dedicated for low power wide area
network. LoRa stack defines mainly MAC and physical layers. The latter
is responsible for the modulation technique that is based on chirp spread
spectrum (CSS) with integrated forward error correction to provide more
interference immunity and higher sensitivity to the receiver [22]. LoRaWAN
defines the MAC layer and provides data rates that range from 250 bps up
to 50 Kbps determined according to the aimed communication range. The
predefined selection of the triple; code rate (CR), spreading factor (SF), and
bandwidth trades off between data rate and communication range [23]. The
prime advantage of LoRa is characterized by its capability to cover up to 15
km and demodulate signals below the noise floor.
12 Chapter 2. Smart Meter Communication

LoRa supports bi-directional communication between base stations and end-

users’ nodes. It operates in the unlicensed ISM band with a spectrum of
920-923 MHz. The chirp frequency band is used to recover the information
signal with a variable bandwidth that can be programmed in the initial set-
tings. A suitable combination of the spreading factor and code rate affects
the coverage area and the end-to-end delay of data packet transfer but at
the expense of the interference immunity. LoRa technology employs the
smart ALOHA technique to save nodes’ energy consumption. It allows the
transmission of packets only when information is needed to be transferred;
otherwise, it prevents it. This asynchronous method extremely enhances
the smart meter lifetime and made LoRa one of the best wireless technolo-
gies for battery-constrained nodes.
Thanks to the star topology and adaptive data rate mechanism that LoRa
uses, it has the ability to optimize the battery lifetime and data rate effi-
ciently. This, in turn, boosts the LoRa capacity and ability to collect plenty
of information from a dense network. LoRa employs AES encryption and
offers the application and network security to ensure data protection and
authentication. However, this technology does not usually meet low de-
lay requirements and does not support high data rates. Furthermore, its
sending capability is restricted by its duty cycle.

2.3.4 Wi-SUN

Wi-SUN is a wireless communication technology designed for smart cities

and IoT applications. It operates basically in frequencies that range from
470 to 928 MHz and covers up to five kilometers. IEEE802.15.4g defines
Wi-SUN’s physical specification characterized mainly by frequency shift
keying (FSK) modulation and data rates from 50 Kbps to 400 Kbps rela-
tive to the frequency of operation. The key standard IEEE802.15.4e spec-
ifies the low energy (LE) consumption MAC, which uses CSMA/CA for
asynchronous transmission. The upper layers of Wi-SUN stack are defined
by IETF profiles under five working groups (WGs), in which the FAN WG
2.3. Wireless metering technologies and standards 13

includes the common RPL protocol in its network layer [24]. For configu-
ration, frame messages are sent between devices, using a predefined trickle
timer to discover neighbors. The timer is identified to manage the fre-
quency of transfer and mitigate the anticipated collision. One of the major
disadvantages of Wi-SUN FAN is the long time taken for the configuration
process, making it a poor candidate for managing communication between
several nodes or smart meters in advanced metering infrastructure.
On the other hand, Wi-SUN technology saves energy and supports suit-
able data rates and multi-hop communication in both uplink and down-
link directions. In addition to its ability to provide efficient propagation
features, it can cover large distances. Therefore, the wireless smart utility
network is a promising technology for hybrid PLC and wireless AMIs as it
achieves low power, low link latency, and high deployment scalability.

2.3.5 Wi-MAX

Worldwide Interoperability for Microwave Access (Wi-MAX) is a wireless

technology that is based on IEEE 802.16. It supports high data rates and a
large area of coverage. The variability of its service options combined with
its QoS features enables it to achieve different smart grid implementation
requirements. A study in [25] and [26] proposed a hybrid wireless WLAN-
Wi-MAX frame for smart metering traffic collection. This structure enables
more extension for network coverage and link quality enhancement. For
smart metering applications, the data is transferred from a specific smart
meter towards the concentrator and collector point using the constructed
mesh topology. The Wi-MAX station is used to carry the information to the
base station.

In fact, Wi-MAX outperforms other wireless technologies in some as-

pects due to its unique features. First, Wi-MAX can cover a distance up to
48 Km, with one BS. For this coverage area, it offers a speed that reaches
70 Mbps. Moreover, the data rate is not highly affected by the channel split
14 Chapter 2. Smart Meter Communication

among multiple customers. Thanks to the built-in quality of service (QoS)

for WI-MAX, it provides the best QoS among other technologies.

2.3.6 LTE

Long Term Evolution (LTE) is a standard wireless technology in the field

of the Internet of Things (IoT). It can be integrated with other technologies
to join the smart metering communication efficiently. LTE offers low delay,
high network capacity, and fast data rate. It also has the ability to enhance
the nodes’ lifetime through adopting OFDMA scheme in the downlink and
single-carrier FDMA in the uplink. LTE supports a variety of modulation
schemes as well as multiple-input multiple-output (MIMO) communica-
tion. It supports a wide bandwidth range of 1.4 to 20 MHz and enables
enough spacing between sub-carriers up to 15 KHz.
The prime challenges of LTE in smart grid and smart metering communi-
cation can be summarised as follows: first, the physical downlink control
channel (PDCCH) is used mainly to manage the scheduling information
of each sub-frame. The number of control information messages will, in
turn, increase in proportion with the number of users. This will finally
cause the exploitation of resources other than the data resources and may
not be enough for a great population of consumers to schedule all of them.
Second, when users are randomly requesting channel access, especially in
dedicated up-link absence, a new problem is initiated. If asynchronous re-
questing is allowed to multiple users, a collision may happen between their
preambles, contributing to higher access latency and lower PDR. In smart
metering, sounding reference signal (SRS) also greatly impacts the whole
system overhead, especially for a high population of end-users. This issue
can be mitigated by enlarging the SRS period or interval as possible even
if there is no active communication. LTE is vulnerable to possible coverage
decline, especially in smart metering, which is due to the node deployment
with very high density. This will, in turn, have a negative influence on the
whole system in terms of interference and coverage. Furthermore, the re-
cent version of LTE is much more expensive than other wireless metering
2.4. Hybrid RF and PLC network 15

technologies.

2.3.7 6LowPAN

IPv6 over Low-Power Wireless Personal Area Network (6LoWPAN) repre-

sents a simple, low-power, and cheap system implementation that offers
packet transmission between devices functioning under IEEE 802.15.4 stan-
dard. It uses a maximum length of IPv6 PPDU of 128 bytes to conform with
IEEE802.15.4 MAC frame length. IPv6 utilizes a header size of 40 bytes.
In the protocol stack of nodes, the adaptation layer serves as the gateway
between 802.15.4 MAC and the IP layer. It is responsible for fragmenting
packets, compressing the header, and then reassembling them. The ZIGBEE
standard’s main goal is the reduction of consumed power to use a simple-
design transceiver. The data rates supported are low, especially in the fre-
quency bands 868 MHz and 915 MHz. Also, it supports a data rate of 250
Kbps in the 2.4 GHz band. On top of that, 6LowPAN keeps up CSMA/CA
for channel access when synchronized nodes use beacons.

2.4 Hybrid RF and PLC network

Recently, most smart metering networks use more than one communica-
tion technology to achieve better system performance. Each of which has
its own specifications characterized by its unique OSI model. While the
PLC technology is characterized by its low deployment cost, it is sensible
to interference and impulse noise that degrades the system reliability. The
wireless radio frequency network may also suffer from noise and interfer-
ence; it offers higher data rate transmissions but with lower reach-ability.
Therefore, utilizing the hybrid of both wireless RF and PLC is useful in
terms of achieving more system reliability. This returns to the fact that there
are no symmetric characteristics of the RF and PLC links.
The hybrid RF and PLC network is recommended as a general solution
to achieve more system reliability and higher coverage distances. However,
this suggested architecture needs further studies and adaptations in order
16 Chapter 2. Smart Meter Communication

to optimize the system performance. The authors in [27] treated the hybrid
network architecture as each technology is separated from the other and has
its own communication section, but this is usually not the real implementa-
tion case. Furthermore, the key to accomplishing more system reliability is
to interconnect the two technologies so each can overcome the drawbacks
of the other.
The current implementation of the hybrid RF and PLC network tends to
use different wireless meter technologies to achieve the system specification
in their vicinity of deployment. They may also be used as repeaters to am-
plify the transmitted signal between a certain node and the destined roots.
Consequently, the resultant network may be a mixture of PLC links that are
characterized by their tree topology architecture and different wireless tech-
nologies such as wireless meter bus, Lora, and Wi-SUN specified by their
mesh connection.

The hybrid of wireless RF and PLC network is mainly affected by the ap-
plication characteristics and the environment they are immersed in. This,
in turn, raised the challenge of combining the hybrid network into a single
infrastructure that is more reliable. Power line communication is mainly
more susceptible to interference and noise and suffers from low data rates,
whereas its short-range characterizes RF network. Hence, using single wire-
less communication will encounter reach-ability challenges, and a single
PLC network will be constrained with its specification. Consequently, re-
cent studies and papers began to suggest different solutions to apply and
implement the hybrid network in real communication to benefit from each
network’s advantages in such a way each can compensate for the draw-
backs of the other.
The RF and PLC networks can both be categorized as Low Power and
Lossy Networks (LLNs). LLN is a group of interconnected embedded de-
vices with limited power, processing, and storage. Such limited network
resources made the selection of an appropriate routing protocol a challeng-
ing task. As the hybrid network architecture exhibits the characteristics
2.4. Hybrid RF and PLC network 17

of LLNs, some constraints and unique routing challenges should be taken

into account. The varied applications mean different characteristics and
conflicting requirements such as latency, reliability, and traffic overhead.
Furthermore, the dominant communication pattern in LLNs is multi-point
to point (MP2P) as well as alternating to other patterns like P2P and P2MP.
Also, the operation of LLNs exhibits deployments with different densities,
i.e., scalability issue, and hence the routing can handle relevant, diverse
cases. The power constraints in LLNs force the routing to ensure low en-
ergy consumption that sends useful updates for the refresh of routes. The
routing protocol that can achieve the LLNs requirements is the RPL; Rout-
ing Protocol for LLNs.

2.4.1 Previously Proposed solutions

There are several routing protocols used for wireless mesh networks. The
routing in power line communication may inherit some of those protocols
to optimize path selection according to its unique specification. For in-
stance, layer 2 of the PRIME PLC MAC is specified by its proactive MAC
routing. [28] revealed comparative analysis of different routing protocols
in wireless mesh networks (WMNs). AODV exhibits the highest delay and
average energy consumption among all. OLSR is characterized by its low
power consumption, packet loss, and average delay. The network charac-
teristics basically control the choice of routing algorithm. For power saving,
PAMAS is a good choice. However, for fault-tolerance purposes, TORA or
ATR can be the optimum path selection algorithms [29]. Few papers com-
pared RPL to other related protocols [30]. The results of [31-32] showed that
RPL outperforms its counterpart LOADng protocol in terms of overhead,
delays, memory, and route formation. [33] also proved that RPL exhibits
lower delay and higher packet delivery ratio (PDR) than AODV.

My proposed solution claims that the system performance optimization

can take place in the network layer to use a routing protocol that works
with both networks like The Routing Protocol for Low Power and Lossy
18 Chapter 2. Smart Meter Communication

Networks (RPL). Recent physical alliance release called wireless smart util-
ity network (Wi-SUN) with its FAN profile contains the RPL in its network
layer and is recommended strongly as a solution for the hybrid network
performance enhancement. [34] suggested using multiple RPL instances in-
spired by existing RPL mechanisms to choose the best interface for routing
per node. The same paper suggests Parent Oriented Design (PO) that aims
to merge the two technologies into a single hybrid metric. Generally, rout-
ing aimed to optimize the path selection in layer 3 without taking into con-
sideration lower OSI layers. Furthermore, the insertion of extra repeaters
in the network environment should contribute to the latency performance
enhancement.
Nevertheless, the architecture of a hybrid network inspires the role of
MAC layer in optimizing the routing performance. Consequently, sug-
gested alternatives introduce the concept of MAC layer extension for both
RF and PRIME technologies. It basically targets applying some modifica-
tions to the RF frame structure without changing MAC components. Car-
rier Sense Multiple Access (CSMA) is applied conventionally in the contention-
free period (CFP), which represents the interval of all guaranteed time slots
of all network nodes. At the same time, RF CSMA is to be applied in the
SCP to replace scheduled channel polling [35]. The major advantage of this
solution is the increase in network reliability and link throughput.
 19

Chapter 3

Realistic wireless smart-meter

network optimization

3.1 Introduction

The smart metering networks like RF-based systems requires a set of fea-
tures that meets the predefined requirements such as latency, energy, and
lifetime. Each energy meter should be able to update its consumption of
gas, water, or electricity in an efficient manner. Hence, a set of requirements
are needed to meet such limitations.

3.2 Routing requirements for the AMI

The smart metering applications along with its unique specifications and
constraints will need a flexible and dynamic routing protocol. In [36], it
is shown that the scalability is one of the most important requirements for
the routing protocol especially when used for LLN. The suitable routing
should not allow the linear and direct proportionality between the routing
scale and the network density. Furthermore, good routing protocol should
be reliable for LLN features. For instance, the link quality should be taken
into account in order to adapt itself with dynamic and fast change of this
metric with time.

For smart grid communication, specially under the umbrella of smart

metering applications, the consumed energy of nodes should be highly
20 Chapter 3. Realistic wireless smart-meter network optimization

considered in the routing selection and topology construction. Therefore,

knowing that most of smart meters are battery-constrained, it is necessary
to enhance and network lifetime and also adapt the routing to adapt to
dynamic variations efficiently. In other words, it should reduce the traffic
overhead of control messages and ensure up-to-date maintenance. On top
of that, the routing for LLNs should avoid loops as possible as to save more
nodes’ power and accelerate more robust topology construction. This can
be achieved through the use of ranks or sequence numbering.
3.3. Routing protocol for low power and lossy networks (RPL) 21

3.3 Routing protocol for low power and lossy networks

(RPL)

In this chapter, we present an overview of the routing protocol for low

power and lossy networks (RPL) and its key role in improving the system
performance of the hybrid wireless and wired network.

3.3.1 Overview of RPL

RPL is a proactive distance vector based routing protocol designed for low
power and lossy networks (LLNs) that are composed of resource constrained
nodes. RPL protocol can operate in both RF and PLC networks and con-
sidered as a solution for networks with limited power, low data rate, and
high packet error rate (PER) [37]. It allows the construction of the topol-
ogy graph named as Destination Oriented Directed Acyclic Graph; which
is a tree rooted at sink without cycles. The tree topology is built based on
predefined objective functions (OFs) that allow the nodes to select the best
parent for forwarding packets based on some routing metrics such as la-
tency, hops, or energy. The main RPL messages as well as the Objective
Function play a key role in forming the DODAG and enabling routing.
This protocol is mainly done through four phases. The first begins with
broadcasting the routing configuration parameters included in the DODAG
Information Object (DIO) message by the root. All the nodes the receive the
DIO message are able to construct their own routing table for future up-
ward or downward communication. The second phase aims to build the
DODAG topology using node rank computation. The rank represents the
location or radius of each node relative to the root and aims to prevent loops
in the constructed tree topology [38]. Similarly, each node sends Destina-
tion Advertisement Object (DAO) to form the shortest path to the DODAG
root. The objective function defines some algorithms to constrain the rout-
ing metrics contained in the received DIO metric container. Each node se-
lects the minimum rank neighbor as its preferred parent for forwarding
packets. DODAG Information Solicitation (DIS) message is used by nodes
22 Chapter 3. Realistic wireless smart-meter network optimization

that want to join the DODAG. RPL protocol uses trickle algorithm to deter-
mine how frequently the DIO messages are sent to enable more DODAG
stability that ensures that constructed topology is up to date and verify en-
ergy efficiency [18].

3.3.2 CONTIKI RPL

RPL routing protocol is commonly used in the IoT community as it does not
impose the use of a specific default objective function and opened the door
to select it according to the defined application and system constraints. One
of the implementations of the RPL is the CONTIKI operating system where
the RPL uses minimum rank with hysteresis objective function (MRHOF)
as the default objective function (OF) with the expected transmission count
(ETX) metric for rank computation. Even for static network nodes, there
may be some dynamic changes in the metric which may cause instabil-
ity in the constructed DODAG. Hence, MRHOF ensures that the selected
preferred parent has the minimum cost while using hysteresis to prevent
excessive churn in the network [39]. ETX is an estimation for the average
number of transmissions to ensure correct delivery to the destination. It is
calculated by actively sending probes between nodes within specified time
window.

The Contiki RPL uses a couple of operating modes. The first storing
mode is appropriate for nodes with constrained storage capacity, as it al-
lows bi-directional communication between source and sink nodes only.
Hence, the whole routing and topology data is known only by the sink
node. That’s why the central sink node is used mainly for routing all nodes.
However, the storing mode tend to allow broadcasting routing and child
nodes’ parameters to contribute to the final DODAG construction. Con-
tiki implementation of RPL allows memory optimization through the em-
ployment of dynamic allocation. Also, it enhances the network energy and
lifetime by switching nodes to sleeping mode in the case of idle events.
A periodic check for the channel is allowed for this manner. Furthermore,
3.3. Routing protocol for low power and lossy networks (RPL) 23

CONTIKI-MAC allows bi-directional communication between network nodes

at low power and with real-time simulation, thanks to the contiki rtimer.
Contiki supports IPv4, IPv6, unicast, broadcast over single and multiple
hops using microIP and rime stacks.

3.3.3 RPL with RF

Due to the recent growth of the Internet of Things (IoT), combined with
the availability of different wireless platforms, RPL has been intensively
studied and evaluated. Smart metering systems may generally require low
latency, and high reliability for most of its applications. Evaluation of RPL
can be divied into two sections: general study and comparison with other
routing protocols, and RPL enhancement approaches to improve its perfor-
mance. In order to evaluate the RPL performance accurately, there should
be a precise emulation and simulation of realistic environment. This neces-
sitates a precise link failure model for attenuation and interference as well
as acquiring a realistic link layer information on real deployment.

Several papers have used the COOJA simulator along with its support-
ing CONTIKI OS to evaluate RPL. A study in [21] have conducted an eval-
uation on randomly generated 86 nodes rooted to single sink node. The
study observed the performance of link quality, end-to-end delay, and traf-
fic overhead. It also reveal the efficiency of RPL with its local repair mecha-
nism and using a single ETX metric for constructing the topology. Further-
more, the authors in [40] proposed some enhancements on RPL specially
on MRHOF and clustering technique. The results show performance en-
hancement of delay, RDR in conjunction with the standard RPL. Also, with
clustering, the consumed energy is reduced. Another study in [41] relates
the network traffic performance in terms of latency with the used RPL ob-
jective function. For example, when MRHOF along with the ETX metric,
the network ability to expand is reduced when compared with OF0 spe-
cially if the number of nodes is greater than 100.
In [42], the observations showed that there is an inverse proportionality
24 Chapter 3. Realistic wireless smart-meter network optimization

between the density of the network and the PDR or stability. The study
also showed that the existence of links with low robustness leads to a high
rate of preferred parent change under RPL, which affects the stability of
DODAG. Another study in [43] focused on the congestion and load balanc-
ing issue at some forwarding nodes. The neighbours of the sink node are
more likely to send their data packets that other nodes in the network. The
congestion mitigation necessitates using a queue utilization factor that in-
cludes multi-casting the congestion data in RPL DIO message. Although
this approach increased the complexity of parent choice, it offers enhance-
ments of network scalability and packet delivery. A study in [44] concerned
with RPL periodic DAO messages and proposed a new approach in order
to dynamically optimize the value of Delay_DAO in accordance to the net-
work density and scale. The optimization can be adapted such that the
rate of constructing DODAG is not slow and also RPL still enable suitable
reaction to DODAG changes. As a result, DAO congestion is mitigated spe-
cially in large networks. Two different studies in [45] and [46] focus on
the effect of used routing metric, under MRHOF, on the stability of the net-
work. Results show that the network stability is reduced under ETX metric.
However, hop-count metric enables more stable network at the expense of
selecting inefficient links.

Two authors in [47] and [48] proposed new enhancements and modifi-
cation on RPL. The first focuses on the use of both artificial intelligence and
cognitive radio in order to optimize the selection of paths from a certain
node to the sink. They used latency and hop metrics in their evaluation of
the AMI scenario. The authors also reveal that cognitive radio contributes
to the improvement of end-to-end delay, consumed energy and PDR but
with extensive alteration of the standard RPL. In [49], the performance of
a new proposed RPL metric is evaluated with ETX metric. The new metric
comprises four path properties which are link quality, link asymmetry, re-
transmission and RDR. The simulation implies the implementation of smart
grid system and the evaluation is conducted for the new metric in both the
3.3. Routing protocol for low power and lossy networks (RPL) 25

existence and absence of ETX. Results reveal better network performance

when the ETX is excluded in terms of the stability. Similarly, the study
in [50] introduce a new RPL OF that aims to combine the effect of several
routing metrics using fuzzy logic. When compared with its MRHOF coun-
terpart, it shows better network lifetime, latency and PDR but at the cost of
more network instability.
In the standard RPL, a trickle timer is adjusted to preserve the DODAG
causing a network overhead. In a way to reduce such overhead, [51] in-
troduce a new Trickle-L2 approach aims to reduce to the network traffic
through the control of the waiting time of nodes for a certain path quality
or the transmission of control messages. For smart grid communication,
nodes may be powered through batteries specially for smart metering ap-
plications like gas and water meters. Therefore, [52] proposed a new RPL
metric that is based on two routing costs; ETX and remaining energy of
nodes. The study demonstrate that network lifetime in improved by 12%.
Another study in this aspect introduce a new RPL extension that takes both
energy and battery index into account. This RPL modification show im-
provement in network consumed energy.

3.3.4 RPL with PLC

In [53], a new modification on RPL is proposed by introducing a new objec-

tive function that aims to suit the rank calculation to the constrains and lim-
itations of power line communication. The simulation environment is con-
figured using real field measurement. The results show better constructed
DODAG at the expense of higher traffic overhead.
The first application of RPL in a PLC medium using communication that is
based on IEEE 802.15.4 ZIGBEE and IPv6 under 6LowPAN was proposed in
[54]. This includes implementation of a real test-bed scenario under CON-
TIKI operating system. This simulation showed better enhancement for the
end-to-end delay by about 45% under ETX metric when compared with the
uncompressed IPv6. In [55], 6 PLC nodes rooted to a single sink node are
26 Chapter 3. Realistic wireless smart-meter network optimization

simulated under COOJA simulator. The simulation results are then com-
pared to a proposed realistic RPL scenario. The outcome confirmed that
the simulation demonstrate lower latency performance than the real sce-
nario due to the increase of induced path layer retries.

Another study in [56] proposed a comparison between RPL and LOAD

routing protocols in G3 power line communication standard. The evalua-
tion is conducted by both realistic test-bed scenario and OPNET simulation.
Results indicate similar performance for both scenarios with higher simu-
lation time for RPL due to its reactive routing.
Furthermore, [57] proposed a channel occupancy metric that takes into con-
sideration different modulation schemes. The study is conducted with the
power line communication IEEE 1901.2 standard. This new metric reveal
higher network stability and enhanced forwarding performance when com-
pared with ETX metric under MRHOF.
On top of that, proposed techniques of polling schemes are introduced
in [58]. This proposal provides an optimization solution in the applica-
tion layer. The automatic reading is tested and evaluated through a simu-
lated PLC environment with one Data Concentrator Unit (DCU) and under
LOADng and RPL protocols. The outcome of the implementation demon-
strate enhancements in the speed of reading as well as the ability to scale
up the network with higher and better reading performance.

3.3.5 RPL simulation

There are different network simulators dedicated for the field of IoT ap-
plications and supporting wireless sensor networks. NS3 and COOJA are
widely used wireless network simulators specially in the field of IOT.
3.3. Routing protocol for low power and lossy networks (RPL) 27

A. Network Simulator 3 - NS3

It is a discrete-event simulator supporting different internet systems and

applications. Building a network with its related routing protocols necessi-
tates adapting existing NS3 modules as well as creating new ones. Specif-
ically, the PRIME PLC network is built by creating new classes, examples,
and tests to emulate the PRIME parameters and related network compo-
nents. However, the wireless RF network can easily be built using existing
models. The implementation of RPL routing protocol needs building a new
module. Since RPL is not supported in the latest release of NS3, devel-
opers suggest a full RPL skeleton that is compatible with NS3 core files.
Since the hybrid network simulation is highly dependent on RPL, the idea
of pursuing with NS3 begins to vanish because the developers expected to
completely implement it within 1 year. COOJA is the alternative network
simulator that implements RPL protocol and offers friendly graphical user
interface (GUI).

B. COOJA - Contiki network simulator

It is based on Contiki operating system and designed for wireless sensor

networks (WSNs) and IoT applications. Cooja provides different simula-
tion mediums each has its unique characteristics, tools, and plugins. The
most common medium is the unit disk graph medium (UDGM) that uses
signal strengths to limit transmission ranges. For our simulation case, in
which signal attenuation and real time environment parameters are given,
Multi-path Ray Tracer medium (MRM) is the best choice. MRM medium
has mainly two plugins: settings, and MRM environment. The settings win-
dow enables to insert all important simulation parameters like background
noise, path loss model,signal attenuation, common transmission power and
so on. The second plugin provides the ability to draw the different types
of attenuation between each pair of nodes. Then, the corresponding Re-
ceived Signal Strength Indicator (RSSI) is demonstrated after the effect of
28 Chapter 3. Realistic wireless smart-meter network optimization

distance, transmission power, and attenuation. In order to enable transmis-

sion power on a node basis, a common sending power is set to all nodes
initially then a node’s push button is used to increase or decrease it. This
way help offer an insight of the minimum and sufficient power that can be
used to achieve the system requirements on a node basis. Unfortunately,
the COOJA simulator does not offer the option of automated transmission
power adjust per node, which should be deemed the optimum.

The proposed composite RPL metric

To achieve the predefined system requirements, especially the latency con-
straint which is 200 milliseconds, the latency is used first as the objective
function. However, the simulation results demonstrate violations of some
network nodes. This is because the delay metric may use undesired paths
and hence increasing the number of re-transmissions and the packet loss
rate causing the final end-to-end delay to exceed the predefined threshold.
Therefore, the proposed Delay-Aware RPL metric (DA-RPL) is introduced.
It is a combination of ETX, delay, and energy. This new additive metric is
given by

DA − RP L = α(ET X) + β(delay) + γ(energy), (3.1)

where α, β, and γ are the relative weights assigned to the three metrics.
Taking into account that the considered metrics are different in their ranges,
the metrics values are normalized to prevent the domination of one specific
metric. For example, ETX metric is normalized to (1 - 1/ETX), and the con-
sumed energy is normalized to 1 - 1/(remaining energy). By manipulating
the coefficients and then choosing the best weights, the final constructed
and converged DODAG topology is expected to achieve the system require-
ments. Hence, the nodes can find paths with high reliability and with lower
energy and delay. The considered energy metric in DA-RPL is the battery
consumption which is accumulated along the simulation time. This means
that the composite metric will not consider it much at the beginning, but
3.3. Routing protocol for low power and lossy networks (RPL) 29

will take it into account more when the battery is consumed with time.
Therefore, it can have a relatively small coefficient and still give reason-
able energy results. As we will see later in the next section, DA-RPL will
achieve the latency requirements and reduce the net consumed enrgy in the
network.

C. Whitefield simulation environment

It is a simulation environment that integrates the simulator NS3 with CON-

TIKI OS COOJA simulator. It extracts the PHY and MAC stack from NS3
and work with NET layer imported from CONTIKI. I used it mainly to
build both wireless and wired PRIME PLC networks separately with the
function RPL routing protocol on the network layer for both. For PLC, RPL
is used to build the tree topology instead of the conventional PRIME rout-
ing that is based on MAC layer. The routing functions under CONTIKI
RPL and used to build the DODAG or tree topology that is directed to a
root or sink without cycles. It uses different costs called Objective Func-
tions (OFs) that optimizes the selection of routes per hop. The final target
to optimize the network performance in terms of end-to-end delay and en-
ergy consumption is to build a fully functional hybrid network that is based
on the comparison between both parallel wireless and wired links in terms
of path cost and congestion. The exploitation of the two technologies indi-
cates the integration or combination of both links through selecting the best
cost among the average of each pair of links.

3.3.6 System model and evaluation results

I. System model

A real wireless smart metering infrastructure is simulated using the CON-

TIKI OS COOJA simulator. This network is located in EL-Minya in Egypt as
shown in Fig. 3.1. The smart meters are emulated with Zolertia Z1 motes
with a compatible cc2420 radio chip. The Multi-path Ray Tracer Medium
30 Chapter 3. Realistic wireless smart-meter network optimization

F IGURE 3.1: El-Minya PRIME network.

(MRM) is used to model the radio environment with its multi-path propa-
gation effects and obstacles attenuation. The transmission model is sched-
uled every defined time interval.
Fig. 3.2 represents the built MRM model where the nodes are injected
and deployed at the same locations of the actual smart meters. To have
better estimates of walls’ attenuation values, communication is established
between two real smart meters in a building and the relative received sig-
nal strength is measured. The obstacles dimensions were estimated from a
given map and inserted into the wireless network. The relative transmission-
reception ratio for each node is affected mainly by distance, obstacles, trans-
mission power, and background noise.
3.3. Routing protocol for low power and lossy networks (RPL) 31

F IGURE 3.2: El-Minya network topology in COOJA

simulator.

CONTIKI-MAC is used for radio duty cycling (RDC) and a channel

check rate of 8 Hz is configured on all nodes to sense the radio activity
without consuming high energy. The data packets of 127 bytes are sent ev-
ery 30 seconds to the sink. The real traffic rate is twice a day or every 12
hours. However, this flow rate was not applied to the simulation because it
will consume too much simulation time. The radio links are assumed to be
asymmetric and the transmission power is configurable on each node sep-
arately. Initially, the power of packet transmission is commonly adjusted to
the maximum, which is 0 dBm. The firmware of z1 motes is changed such
that the transmission power can be lowered with a push-button during the
simulation run. Taking into account the predefined system constraints on
the end-to-end delay as well as the limitations on energy consumption, both
performance parameters are calculated and evaluated. Table I presents the
used simulation environment parameters.
32 Chapter 3. Realistic wireless smart-meter network optimization

TABLE 3.1: Simulation parameters

Parameters Value
Simulation tool Contiki/COOJA 2.7

PHY and MAC layer IEEE 802.15.4

Network layer RPL/DA-RPL

Adaption layer 6LowPAN

Number of nodes 35

Carrier frequency 868 MHz

Data rate 250 Kbps

Data rate transfer 127 byte payload every 30 seconds

II. Performance evaluation

To evaluate the performance of the proposed DA-RPL on El-Minya net-

work, we used the COOJA simulator with the CONTIKI operating system.
The main considered parameters in measurement are the end-to-end delay
and energy consumption. Multi-path Ray Tracer (MRM) is the used radio
model, wherein the obstacles are inserted to emulate real buildings attenu-
ation. EL-Minya network composed of one Data Concentrator Unit (DCU)
and smart meters located mostly on the ground floor. At the node level,
the predefined constraint is a latency of 200 milliseconds to collect the data
within an acceptable time. The results are considered when reaching the
RPL convergence state where each node has a stable parent and a low CPU
power. The threshold to change the preferred parent is adjusted to a suit-
able value to improve RPL stability.

A. End-to-End Delay
End-to-end delay is defined as the time taken to transmit a data packet from
a certain node to the sink. After several simulations, and using a single RPL
metric as the objective function, the latency constraint is not achieved. The
 3.3. Routing protocol for low power and lossy networks (RPL) 33

600

End-to-end delay (ms)

500

400

300

200

100

0
8 26 31
Node ID

ETX Latency Energy DA-RPL with α=0.5, β=0.4, ϒ=0.1

F IGURE 3.3: Nodes end-to-end delay.

second attempt is using the delay as a constraint rather than a metric and
RPL tend to select paths that only meet the constraint and nullify or re-
move other parents. However, this way has not succeeded due to packet
re-transmissions and congested nodes. Since Expected Transmission Count
(ETX) defines the link reliability, it seems a good candidate to avoid conges-
tion. Therefore ETX takes the lead of lowering the node latency as possible.
Having the trade-off between energy consumption and the latency, multi-
ple additive RPL metrics are used to save the power of the node subject to
the latency limitation. The current real PRIME PLC network has a reacha-
bility percentage of almost 80 %. After simulating its RF counterpart net-
work, the modified RPL DA-RPL with suitable metrics weights, and after
the insertion of extra nodes the reachability is improved more. The extra
smart meters are injected to achieve more reliable communication coverage
and reach better latency results. By focusing on the nodes with high cal-
culated latency, Fig. 3.3 shows nodes’ end-to-end delay for four different
routing metrics with respect to the threshold. After manipulating the met-
rics weights, DA-RPL shows the best delay performance and succeeded to
improve the system requirement with α = 0.5, β = 0.4, and γ = 0.1.
34 Chapter 3. Realistic wireless smart-meter network optimization

Repeaters placement algorithm

In the previous RPL simulation section, we have proposed a new com-
posite routing metric to achieve the target latency constraints. However, in
many scenarios and as will be seen in the simulation section, the new com-
posite metric can result in performance gain but it may not be able to attain
the desired latency constraint. We propose to place some repeater nodes
in the network to enhance the latency performance. These nodes will not
generate traffic over the network but can be used to route some traffic. The
placement of these nodes is based on the observation that the cause of la-
tency violation is that some nodes in the network will be congested as they
will route a lot of packets. We place our repeaters to relieve these congested
nodes and to provide some alternative routes to the DCU. The core advan-
tage of this algorithm is the reduction of network congestion and nodes’
E2E delay.

The extra nodes or repeaters are then inserted to the simulation environ-
ment as to enhance the latency performance further. They are injected near
the concentrated motes to mitigate the congestion and also in the vicinity
of high-delay nodes in order to be their default next hop to sink. Fig. 3.4
represents the modified network topology after the insertion of the extra
repeaters.
The same DA-RPL metrics coefficients are used again to check the E2E
delay of nodes relative to the specified threshold. As shown in Fig. 3.5, the
nodes with highest E2E delay finally achieved a latency below 200 msec.
Furthermore, the average latency performance of the whole network is slightly
improved after the insertion, as demonstrated in Fig. 3.6. This means that
each node has a reasonable calculated latency relative to the constraint.
By taking into consideration the stationary nature of El-Minya smart me-
ters’ network, the new RPL modification as well as the insertion of extra
repeaters should provide a reliable, long-term solution for achieving the
target latency constraints.
 3.3. Routing protocol for low power and lossy networks (RPL) 35

F IGURE 3.4: El-Minya network topology after the insertion

of extra nodes.

350
End-to-end delay (ms)

300
250
200
150
100
50
0
5 8 26 31
Node ID

ETX Latency Energy DA-RPL with α=0.5, β=0.4, ϒ=0.1

F IGURE 3.5: Nodes end-to-end delay.

36 Chapter 3. Realistic wireless smart-meter network optimization

250

Average E2E Delay (ms)

200

150

100 After insertion

Before insertion
50

0
ETX Latency Energy DA-RPL
Routing metric

F IGURE 3.6: Network average end-to-end delay.

B. Energy Consumption
Although the energy metric is assigned the lowest weight in DA-RPL, it
seems to give reasonable power consumption results because it is additive
along the simulation time. For different network infrastructure, new appro-
priate weights can be used to optimize the system performance and meet
some predefined specifications. The energy consumption of El-Minya sim-
ulated network is calculated after the insertion of extra repeaters using the
CONTIKI power tracer. The calculated values are taken after the conver-
gence of the constructed DODAG and the traffic flow of control messages
begins to deteriorate. The average power consumption of some network
nodes is shown in Fig. 3.7. It can be clearly confirmed that consumed power
is reduced to the minimum when DA-RPL is used with suitable weights.
Moreover, Fig. 3.8 shows that average power consumption of the whole
network saved the best energy when compared to other routing metrics.
These results show an evaluation for EL-Minya case study.
From the above, it can be seen that the modified RPL, with the selection
of suitable weights, succeeded to achieve lower average consumed power
than using the ETX and latency metrics (as clearly, DA-RPL has energy as
part of the routing metric which is not the case for ETX and latency met-
rics). It also shows power consumption near that achieved by the energy
metric. Since El-Minya network is characterized by its constant bit rate,
3.3. Routing protocol for low power and lossy networks (RPL) 37

Average power consumption (mW) 3.5

3
2.5
2
1.5
1
0.5
0
5 12 26 29 36
Node ID

ETX Latency Energy DA-RPL with α= 0.5,β= 0.4,andγ= 0.1

F IGURE 3.7: Nodes average power consumption.

2.5
Average network power
consumption (mW)

1.5

0.5

0
ETX Latency Energy DA-RPL with
α=0.5, β=0.4,
ϒ=0.1
Routing metric

F IGURE 3.8: Network average power consumption for four

different routing metrics.
38 Chapter 3. Realistic wireless smart-meter network optimization

static nodes, and static environment, the chosen weights can be used indef-
initely unless some major changes occur. In this case, the weights should
be dynamically manipulated to keep up with the changes and to ensure
acceptable performance (in terms of latency and/or power consumption).
Therefore, the chosen RPL weights along with the placed repeater nodes
can provide an effective means for meeting the required latency constraint
while achieving some energy consumption gains.
 39

Chapter 4

Asynchronous Transmission of
Data Packets

4.1 Introduction

This chapter’s main idea aims to present newly proposed algorithms for
data packet transmission to improve network latency performance further.
The delay enhancement is based on adjusting the beginning time of data
frame transmission that ensures less congestion with other meters’ frames.
This chapter focuses on the methods and algorithms used to optimize
the network performance in terms of end-to-end delay using ordered asyn-
chronous transmission of packets. The proposed algorithms are based on
classifying the nodes according to their given synchronous E2E delay. They
are classified into high, medium, and low delay nodes as the first step of
filtration. The second step aims to divide nodes into groups according to
their forwarding nodes or preferred parents to the root or sink. The for-
warding node or parent is determined through the RPL routing protocol
and its corresponding routing tables on a node basis.

Asynchronous implementation is important for many smart grid com-

munications, especially smart metering applications, to enhance network
performance. This solution is analyzed as the realistic RF or PLC scenario
offers such an option for system performance optimization. The ability of
smart meters to transmit this reading information in different starting times
40 Chapter 4. Asynchronous Transmission of Data Packets

enables an intensive enhancement for the end-to-end delay, offers faster

reading, and avoid delays by considering that latency enhancement can be
achieved using the network layer solution like RPL routing protocol. The
system performance can be improved in different aspects by manipulating
different routing metrics such as ETX, delay, hop count, and energy and
their corresponding weight. However, the asynchronous transmission al-
gorithm is considered a contribution for further system enhancement on
top of the routing solution.

4.2 Proposed algorithms

4.2.1 The rationale for proposing new algorithms

The main reason to introduce new algorithms for data packet transmission
is to enhance the system’s end-to-end delay performance further. The la-
tency improvement is based on controlling the starting time of data packet
transmission that ensures less collision with other nodes’ packets. The algo-
rithms’ results are compared with both the synchronous and random asyn-
chronous data transmission cases. Because asynchronous transmission al-
ways outperforms synchronized transmissions, it is important to propose
automated asynchronous criteria that show greater results than the ran-
domized case. The realistic scenario of smart metering applications usu-
ally supports individual configuration of each smart meter in the network.
Hence, we can exploit this option to adjust the beginning sending time of
each node’s UDP packet and mitigate the network congestion.

4.2.2 Algorithm 1

This algorithm aims to classify nodes according to their given synchronized

end-to-end delay. The nodes with the highest delay will then be assigned
enough time guard before and after their data packet transmission. How-
ever, the nodes with lower delay will be sent asynchronously with a time
guard relative to their synchronous delay. In other words, the greater the
nodes’ end-to-end delay in the synchronous scenario, the more the guard
4.2. Proposed algorithms 41

time band before and after the data packet transmission. This approach re-
sults in reduced node latency, as we will see later in the next results section.

Algorithm 1:
1 for Every node do
2 Classify into a delay category (high medium low)
if E2Edelay > 200ms then
3 Set as high delay nodes

4 100ms < E2Edelay < 200ms

Set as medium delay nodes

5 E2Edelay < 100ms

Set as low delay nodes
for Each delay category do
6 Add enough time guard before and after high delay nodes.
Allow asynchronized transmission of the remaining nodes

4.2.3 Algorithm 2

The network nodes go through two main filtration steps. The first is the
delay classification, in which the nodes are divided into high, medium, and
low delay nodes. The latency statistics are given from the synchronous
scenario. The second filtration is to group nodes according to their common
forwarding nodes (CFNs).
The routing table or DODAG topology constructed through the syn-
chronous transfer of UDP packets is used to realize and divide the nodes
according to their forwarding nodes. The nodes with the same preferred
parent are assigned to a single group. The high-delay nodes will be as-
signed a suitable guard time before and after transmission. Also, enough
time separation will be assigned to nodes with CFN. However, nodes with-
out CFN will allow synchronous transmission of their data packets. This is
because these nodes will never collide with each other, and the interference
is unlikely to occur.
42 Chapter 4. Asynchronous Transmission of Data Packets

Algorithm 2:
1 for Every node do
2 Classify into a delay category (high medium low)
if E2Edelay > 200ms then
3 Set as high delay nodes

4 100ms < E2Edelay < 200ms

Set as medium delay nodes

5 E2Edelay < 100ms

Set as low delay nodes
6 for Each delay category do
7 Filter the nodes with Common Forwarding Node (CFN)
Add enough time guard before and after high delay nodes
Allow transmission for nodes with CFN according to their
synchronous delay.
Allow synchronized transmission of nodes without CFN.

4.3 Comparison between RF and PLC

A comparison is conducted between wireless RF and PLC PRIME networks

separately in terms of nodes’ end-to-end delay using both algorithms. Re-
sults show that the delay can be reduced for all network nodes below the
predefined threshold by using just a single RPL routing metric (ETX).

4.3.1 Wireless RF network

For the realistic wireless RF network located in El Minya, the smart meters
are simulated using the Whitefield simulation environment. The nodes are
inserted in the same location as the realistic scenario. A suitable link failure
or path loss model is used with a path loss exponent of 5.7 as to emulate
the realistic attenuation and the existence of obstacles and buildings. The
links between nodes are assumed to be asymmetric such that the RSSI is the
same for any two end-devices.
The smart meters are emulated with sky motes with a compatible cc2420
radio chip. This driver functions as the alternative for the actual cc1312
chip and offers a transmission power range between -10 dBm to 0 dBm.
4.3. Comparison between RF and PLC 43

For radio duty cycling (RDC), CONTIKI-MAC is used along with a

channel sense rate of 8 Hz as the main nodes’ configuration. The user data-
gram protocol (UDP) data packets are sent at a rate of 30 seconds to the
sink node with a size of 127 bytes in order to conform with the 6LowPAN
standard. Although reading data’s realistic traffic rate is repeated every 12
hours or twice a day, this rate is not applied to the Whitefield simulation
environment to save too much simulation time. Hence, the results can be
reliable because it will give a good insight for longer traffic interval. The
radio links are assumed to be asymmetric, and the transmission power is
configurable on each node separately. The transmission power is adjustable
on a node basis, and it is commonly adjusted to 0 dBm.

Async Sync Algorithm 1 Algorithm 2

400

300
E2E delay (ms)

200

100

0
1

9
10

12

13

14

18

19

20

21

22

27

28

29

30

31

32

33

ve 4
ge
3
ra
Node ID A

F IGURE 4.1: End-to-end delay comparison for RF network

between synchronous, random asynchronous, and both
algorithms

Fig. 4.1 shows the end-to-end delay of the RF nodes for different syn-
chronization algorithms. ETX is used as a single routing RPL metric, and it
is selected to offer better latency results as it is considered the best candidate
for link reliability and congestion mitigation. Other routing metrics may
also be considered for other system requirements like energy consump-
tion and network lifetime. Whitefield environment support all Contiki RPL
routing metrics under minimum rank with hysteresis objective function. It
supports hop count, delay, and energy metrics. Also, the objective function
44 Chapter 4. Asynchronous Transmission of Data Packets

may be modified for better performance results. However, for our case,
which requires a minimum end-to-end delay, it is preferred to use the ex-
pected transmission count for path costs. The value of ETX is updated with
each DIO message broadcast between nodes and updated for nodes’ rout-
ing table.

On top of the routing solution to improve the delay, the previously men-
tioned composite RPL metric is excluded, and the asynchronous algorithms
are the alternative. This idea’s emergence stems from the fact that most
smart meters in the real scenario can be controlled through its starting trans-
mission time of data packets. In other words, the beginning of UDP data
transmission is adjustable through all smart meters in the network. That’s
why it is beneficial to further exploit such a feature to further enhance net-
work performance, especially in terms of delay and energy consumption.
The smart meters in the real scenario includes gas and water meters as well
as electric meters for aggregation and collection of readings. A predefined
E2E delay of 200 milliseconds is needed on a node basis. Both synchronous
and asynchronous delay statistics are used to compare for new defined al-
gorithms 1 and 2. The main goal is to achieve better delay results than the
random asynchronous case.

Async Sync Algorithm 1 Algorithm 2

400

300
E2E delay (ms)

200

100

0
7 8 13 18 27 34

Node ID

F IGURE 4.2: E2E delay performance comparison for some

RF network nodes
4.3. Comparison between RF and PLC 45

Algorithm 1 shows better delay results for some nodes and worse la-
tency performance for others. Nevertheless, the whole system’s average
net delay is almost the same as the random asynchronous case, with all
nodes having a delay below the prespecified threshold. As we can see in
the figure above, the end-to-end delay is improved. The used algorithm 2
succeeded in achieving the predefined latency requirement without the use
of the composite additive RPL metric. Therefore, algorithm 2 demonstrates
improved overall latency performance with the best system averaged de-
lay.
For future insights, the transmission of UDP data packets can be managed
automatically and in dynamic behavior. This necessitates a periodic broad-
cast of synchronous end to end delay through the whole network. This
information may be included in the control messages, which enables divid-
ing nodes into three delay categories; high, medium, and low. Moreover,
each node can configure itself for an instant time for transmission. In order
to save power consumption of the network, the smart meters may enter the
sleep mode during the idle state while maintaining the bidirectional com-
munication of broadcast messages like DIO and keep alive. This will ensure
that all nodes in the network are joined and synchronized to the DODAG
and keep multi-hopping efficient.

4.3.2 PRIME PLC network

For the realistic PRIME PLC network located in El-Minya, the smart me-
ters are simulated using the Whitefield simulation environment. Two types
of cables are mainly used to emulate the real power line communication,
underground and overhead cables. The cables are mainly characterized by
their transmission and reception interface impedance and attenuation and
interference.

A suitable transmission power spectral density (PSD) is used for the

simulation that conforms with the PRIME PLC standard. The current num-
ber of repeaters are actually inserted into the simulation environment to test
46 Chapter 4. Asynchronous Transmission of Data Packets

the coverage, delay, and energy performance. The default topology forma-
tion in PRIME takes place in the MAC layer using layer 2 routing. The
topology construction is based on the shortest path algorithm at which the
DCU is considered the base node (BN) and others are service nodes (SN).
However, this routing construction is exchanged with the RPL routing pro-
tocol in layer 3 (network layer). In other words, PRIME MAC L2 routing
is exchanged with RPL routing protocol. The front end of PLC nodes is
modified such that it interprets and understands radio frequency wireless
communication. An interface of cc2420 driver is used to do this task.
This is to give insight into future adoption and adaptation of G3 PLC
standard, enabling RPL in its stack. To have a good insight into future hy-
brid solution, it is important to study RF and PLC’s performance separately
and then apply the combined network accordingly. CONTIKI-MAC is the
layer employed for duty cycling of radio channels with a configurable link
check rate of 8 Hz. The User Data-gram Protocol (UDP) is managed by
the transport layer in which the data packet transmission rate is set to 30
seconds. The packet size is adjusted to 127 bytes, which is the maximum
length for the 6LowPAN standard.

F IGURE 4.3: End-to-end delay comparison for PLC

network between synchronous, random asynchronous,
and both algorithms

The UDP packets are mainly unicasted from each node to the destined
root by defining the client and server port numbers. The sink or DCU, in
turn, waits for packets from smart meters, receives, and finally prints them.
As seen in Fig. 4.3, the PRIME PLC network’s end-to-end delay shows
4.3. Comparison between RF and PLC 47

TABLE 4.1: Simulation parameters

Parameters Value
Simulation tool Whitefield environment

PHY and MAC layer IEEE 802.15.4

Network layer RPL

Adaption layer 6LowPAN

Objective function MRHOF with ETX metric

Transmission PSD -50 dBm/Hz

Data rate 250 Kbps

Data rate transfer 127 byte payload every 30 seconds

worse than its wireless RF counterpart.

For the current PRIME model applied in the Whitefield, we should con-
sider that RPL shows poor delay results when applied to the whole network
framework. Consequently, RPL is applied and implemented to different in-
dividual sections in the PRIME network. Hence, we have multiple sectors
of smart meters configured by RPL such that the UDP sink server is com-
mon among all sectors. The repeaters are inserted in the same location as
the realistic scenario t ensure 100 % area of signal coverage.

F IGURE 4.4: E2E delay performance comparison for some

PLC network nodes
48 Chapter 4. Asynchronous Transmission of Data Packets

In fact, node 27 is the farthest meter from the root or data concentrator
unit. This is the main reason for its high delay performance. Node 7, how-
ever, uses multiple hops to reach the sink destination. Generally, two cables
are used to emulate the realistic scenario, the underground and overhead
cables. Generally, the overhead cables show high signal attenuation, which
is greater than the one of the underground cables. This is mainly due to the
path loss in the air beside the existence of obstacles and buildings that may
cause interference and signal reflection.

4.3.3 RF against PLC performance

It should be noted that the delay curve for RF is plotted with milliseconds,
whereas the PRIME PLC curve reveals delay calculated in seconds. This
study confirms that wireless RF under RPL and asynchronous transmission
outperforms the PRIME PLC network with the same configuration. This
may give an insight for future adaptation to the hybrid model that most of
the RF links may be selected instead of the PLC links to achieve such de-
lay predefined requirements. For the general case, each node may have an
equal probability of choosing either RF or PLC according to the configured
path cost or routing metric and its relation to the system requirements. For
example, if the major system requirement is to minimize the delay for E2E
communication, the technology with lower delay cost will be selected. Sim-
ilarly, if the main network concern is link reliability and higher PDR, then
the PLC technology will mostly be chosen.

Finally, a suitable automated and dynamically adjusted transmission

mechanism can be used to ensure successful packet reception for all nodes
without causing interference or congestion in the whole network. This can
be achieved through machine learning algorithms that need plenty of his-
torical data of the network performance and offer a near-accurate model
for the network. The last approach may optimize the overall system energy
consumption and increase the nodes’ lifetime through precise and dynamic
power configuration.
 49

Chapter 5

Conclusions and Future Work

5.1 Conclusions

In this work, we introduced a new RPL routing metric DA-RPL to reduce

the nodes’ end-to-end delay below a specified threshold by selecting paths
with less latency and reduced congestion. Extra nodes were inserted as
repeaters to contribute to latency reduction and to reduce the congestion
at concentrated nodes. The results reveal that all system nodes met the
predefined requirement and with less consumed energy. To generalize for
different network environments, the coefficients of the metric can be manip-
ulated dynamically with the changes in the network to optimize the system
specifications.

Furthermore, the new proposed algorithms of asynchronous transmis-

sion combined with appropriate ordering show better network performance
results in terms of end-to-end delay as well as consumed energy among
nodes. The significant reduction in the nodes’ latency is mainly due to effi-
cient congestion in the whole network by avoiding packet collision. Results
also confirmed that RF wireless network may reveal better performance
than PRIME PLC network, specially if the attenuation and obstacles be-
tween nodes are not high relative to the transmission power.
50 Chapter 5. Conclusions and Future Work

5.2 Future directions

For future work, appropriate automated transmission power mechanism

beside the composite metric can be used to provide better congestion miti-
gation and offer a promising enhancement in the network performance in
terms of delay and energy consumption. Furthermore, a wired PLC net-
work can be combined with a wireless system and a modified RPL with a
hybrid metric can be used to enhance the system performance.
Several research studies are done on hybrid modeling in order to imple-
ment it in the realistic case. An adaptation to the G3-standard can emerge
the possibility for hybrid routing under RPL umbrella by using the net-
work layer of the G3-PLC. Furthermore, the new wireless technology, Wi-
SUN, can utilize its RPL protocol in its stack to integrate with G3 stack. The
hybrid routing metric can be identified independently on each of the two
technologies according to its requirements and challenges with its links.
There are three different proposed methods that can be used to integrate
both wireless and wired technologies under RPL routing. All of them as-
sumes that all nodes should have two interfaces; one for RF (802.15.4) and
one for PLC (PRIME).
The first method is called Multiple RPL Instances (MI) that aims to use two
RPL instances one per technology. For each node interface, specific objec-
tive function OF, and instance id are identified. Hence, the net is two con-
structed DODAGs rooted to a single sink node. The primary technology
is assigned a lower RPL instance ID, whereas the backup technology is de-
fined by a higher instance ID. An appropriate switching between the two
interfaces occurs upon failure. Since RPL only enables transition to a higher
instance ID, the backup possibility is not high. However, MI method can
be easily implemented and offers high stability to the network. In order to
have more advantage from the multiple interface feature, the limitation on
switching between different instances should be altered. The new proposed
concept should take into consideration the looping issue as well as the num-
ber of switches between interfaces or technologies. It may enable the nodes
5.2. Future directions 51

to access or join more than one instance to reach its sink node through va-
rieties of selected routes’ choices. According to the predefined system re-
quirements, the route metrics, and the link capacity, each node should be
able to choose the optimum instance for forwarding its data packets to the
root.
For far future, it is important to take into account the long range solution
in the hybrid wireless and PLC network despite its restricted bandwidth.
The multiple interface management can be controlled through using an al-
gorithm that allows direct link from smart meter to the sink node. This ded-
icated link should indicate long distance communication with new adapta-
tions to the routing protocol with a specific routing metric. This mechanism
is expected to offer availability for critical packet transmission with lowest
delays.
52

References

[1] O. A. Gonzalez, J. Urminsky, M. Calvo and L. de Haro, “Performance

analysis of hybrid broadband access technologies using PLC and wi-fi,”
in Proc. International Conference on Wireless Networks, Communications
and Mobile Computing, 2005, pp. 564-569 vol.1.
[2] F. Entezami, C. Politis, "Routing protocol metrics for wireless MESH net-
works", Wireless World Research Forum, 2013
[3] Eiman Alotaibi and Biswanath Mukherjee. 2012. Survey paper: A sur-
vey on routing algorithms for wireless Ad-Hoc and mesh networks. Com-
put. Netw. 56, 2 (February, 2012), 940–965.
[4] Abdel Hakeem, S.A.; Hady, A.A.; Kim, H. RPL Routing Protocol Perfor-
mance in Smart Grid Applications Based Wireless Sensors: Experimental
and Simulated Analysis. Electronics 2019, 8, 186.
[5] La, C.A.; Heusse, M.; Grenoble, A.D. Link reversal and reactive routing
in Low Power and Lossy Networks. In Proceedings of the 2013 IEEE 24th
International Symposium on Personal Indoor and Mobile Radio Commu-
nications (PIMRC), London, UK, 8–11 September 2013; pp. 3386–3390.
[6] Iyer, G.; Agrawal, P.; Cardozo, R.S. Performance comparison of routing
protocols over smart utility networks: A simulation study. In Proceedings
of the 2013 IEEE Globecom Workshops (GC Wkshps, Atlanta, GA, USA,
9–13 December 2013; pp. 969–973.
[7] Kathuria, V.; Mohanasundaram, G.; Das, S.R. A simulation study of
routing protocols for smart meter networks. In Proceedings of the 2013
IEEE International Conference on Smart Grid Communications (SmartGrid-
Comm), Vancouver, BC, Canada, 21–24 Octember 2013; pp. 384–389.
[8] F. Lemercier, N. Montavont, L. Toutain, K. Vijayasankar, R. Vedantham,
P. Chiummiento, "Support for hybrid network in RPL", Proc. IEEE Smart-
GridComm, pp. 527-532, 2016.
[9] L. Sr’astny, L. Franek, P. Fiedler, "Wireless communications in smart me-
tering", 12th IFAC Conference on Programmable Devices and Embedded
Systems, pp. 330-335, 2013.
5.2. Future directions 53

[10] K. Zeman, P. Masek, J. Krejci, A. Ometov, J. Hosek, S. Andreev, F.

Kroepfl, "Wireless M-BUS in Industrial IoT: Technology Overview and Pro-
totype Implementation", European Wireless 2017; 23th European Wireless
Conference; Proceedings of, pp. 1-6, 2017.
[11] C.M. Ramya, M. Shanmugaraj, R. Prabakaran, "Study on ZigBee tech-
nology", 3rd International Conference on Electronics Computer Technology
(ICECT), pp. 297-301, 2011.
[12] J. Malhotra Manpreet, "ZigBee technology: Current status and future
scope", 2015 International Conference on Computer and Computational
Sciences (ICCCS), pp. 163-169, 2015.
[13] Umber Noreen, Ahcne Bounceur, Laurent Clavier, "A Study of LoRa
Low Power and Wide Area Network Technology", 3rd IEEE International
Conference on Advanced Technologies for Signal and Image Processing
(ATSIP’2017), May 2017.
[14] M. Saari, A. M. bin Baharudin, P. Sillberg, S. Hyrynsalmi, W. Yan, "Lora
- a survey of recent research trends", 41st International Convention on In-
formation and Communication Technology Electronics and Microelectron-
ics (MIPRO) Opatija HR, May 2018.
[15] C.-S. Sum, M. A. Rahman, L. Lu, F. Kojima, H. Harada, "On communi-
cation and interference range of ieee 802.15. 4g smart utility networks",
Proc. 2012 IEEE Wireless Communications and Networking Conference
(WCNC), pp. 1169-1174, 2012.
[16] O. Iova, G. P. Picco, T. Istomin, C. Kiraly, "RPL: The routing standard
for the Internet of Things. . . Or is it?", IEEE Commun. Mag., vol. 54, no. 12,
pp. 16-22, Dec. 2016.
[17] T. Junjalearnvong, T. Habara, R. Okumura, K. Mizutani, H. Harada, "A
Dynamic Routing Protocol Supporting Mobile Nodes in Wi-SUN FAN Sys-
tems", Proc. WPMC 2017, pp. 325-331, December 2017.
[18] N. Sousa, J. V. V. Sobral, J. J. P. C. Rodrigues, R. A. L. Rabêlo, P. Solic,
"ERAOF: A new RPL protocol objective function for Internet of Things ap-
plications", 2017 2nd International Multidisciplinary Conference on Com-
puter and Energy Science (SpliTech), pp. 1-5, 2017.
54 Chapter 5. Conclusions and Future Work

[19] Nguyen Thanh Long, N. De Caro, W. Colitti, A. Touhafi and K. Steen-

haut, "Comparative performance study of RPL in Wireless Sensor Networks,"
2012 19th IEEE Symposium on Communications and Vehicular Technology
in the Benelux (SCVT), Eindhoven, pp. 1-6, 2012.
[20] Renesas 2019, Sub-GHz Wireless Communication Solutions,
https://www.renesas.com/eu/en/solutions/proposal/subghz.htmloverview
[21] Francois Lemercier, G. Habault, G. Z. Papadopoulous, P. Maille, P.
Chatzimisios, and N. Montavont. “Communication architectures and tech-
nologies for advanced Smart Grid Services.” In: Transportation and Power
Grid in Smart Cities: Communication Networks and Services. UK: John
Wiley, 2017. Chap. 8.
[22] Lemercier F., Montavont N. (2018) Performance Evaluation of a RPL
Hybrid Objective Function for the Smart Grid Network. In: Montavont N.,
Papadopoulos G. (eds) Ad-hoc, Mobile, and Wireless Networks. ADHOC-
NOW 2018. Lecture Notes in Computer Science, vol 11104. Springer, Cham.
https://doi.org/10.1007/978-3-030-00247-3_3
[23] F. Lemercier, N. Montavont, L. Toutain and P. Chiummiento, "A New
Objective Function for Hybrid Network in the Smart Grid," 2018 IEEE 19th
International Symposium on "A World of Wireless, Mobile and Multime-
dia Networks" (WoWMoM), Chania, 2018, pp. 14-16, doi: 10.1109/WoW-
MoM.2018.8449744.
[24] F. Lemercier, N. Montavont, L. Toutain, K. Vijayasankar, R. Vedantham
and P. Chiummiento, "Support for hybrid network in RPL," 2016 IEEE Inter-
national Conference on Smart Grid Communications (SmartGridComm),
Sydney, NSW, 2016, pp. 527-532, doi: 10.1109/SmartGridComm.2016.7778815.
[25] T. Ropitault, A. Pelov, L. Toutain, R. Vedantham and P. Chiummiento,
"Channel occupancy - Faster, better, stronger," 2015 IEEE International Con-
ference on Smart Grid Communications (SmartGridComm), Miami, FL, 2015,
pp. 695-700, doi: 10.1109/SmartGridComm.2015.7436382.
[26] J. R. R. Renofio, M. E. Pellenz, E. Jamhour, A. Santin, M. C. Penna and
R. D. Souza, "On the dynamics of the RPL protocol in AMI networks under
jamming attacks," 2016 IEEE International Conference on Communications
5.2. Future directions 55

(ICC), Kuala Lumpur, 2016, pp. 1-6, doi: 10.1109/ICC.2016.7511150.

[27] T. Ropitault, A. Lampropulos, R. Vedantham and P. Chiummiento,
"Realistic model for narrowband PLC for advanced metering infrastruc-
ture," 2013 IEEE International Conference on Smart Grid Communications
(SmartGridComm), Vancouver, BC, 2013, pp. 624-629, doi: 10.1109/Smart-
GridComm.2013.6688028.
[28] J. R. R. Renofio, M. E. Pellenz, E. Jamhour, A. Santin, M. C. Penna and
R. D. Souza, "On the dynamics of the RPL protocol in AMI networks under
jamming attacks," 2016 IEEE International Conference on Communications
(ICC), Kuala Lumpur, 2016, pp. 1-6, doi: 10.1109/ICC.2016.7511150.
[29] T. Ropitault, A. Lampropulos, A. Pelov, L. Toutain, R. Vedantham and P.
Chiummiento, "Doing it right — Recommendations for RPL in PLC-based
networks for the Smart Grid," 2014 IEEE International Conference on Smart
Grid Communications (SmartGridComm), Venice, 2014, pp. 452-457, doi:
10.1109/SmartGridComm.2014.7007688.
[30] Vibha Kathuria, Gokulraj Mohanasundaram and Samir R Das, "A Sim-
ulation Study of Routing Protocols for Smart Meter Networks", pp. 384-389,
2013.
[31] Nico Saputro, Kemal Akkaya, and Suleyman Uludag. “A survey of
routing protocols for smart grid communications.” In: Computer Networks
56.11 (2012), pp. 2742–2771. doi: 10.1016/j.comnet.2012.03.027.
[32] Z. Yang, S. Ping, A. Aijaz and A. Aghvami, "A Global Optimization-
Based Routing Protocol for Cognitive-Radio-Enabled Smart Grid AMI Net-
works," in IEEE Systems Journal, vol. 12, no. 1, pp. 1015-1023, March 2018,
doi: 10.1109/JSYST.2016.2580616.
[33] Z. Yang, S. Ping, H. Sun and A. -H. Aghvami, "CRB-RPL: A Receiver-
Based Routing Protocol for Communications in Cognitive Radio Enabled
Smart Grid," in IEEE Transactions on Vehicular Technology, vol. 66, no. 7,
pp. 5985-5994, July 2017, doi: 10.1109/TVT.2016.2617874.
[34] JP. Vasseur, M. Kim, K. Pister, N. Dejean and D. Barthel, "Routing Met-
rics Used for Path Calculation in Low-Power and Lossy Networks", RFC
6551 (Proposed Standard), March 2012.
56 Chapter 5. Conclusions and Future Work

[35] Chauvenet, B. Tourancheau, D. Genon-Catalot, P.-E. Goudet and M.

Pouillot, "A Communication Stack over PLC for Multi Physical Layer IPv6
Networking", Smart Grid Communications (SmartGridComm) 2010 First
IEEE International Conference on, pp. 0-5, 2010.
[36] M. Zimmermann and K. Dostert, "Analysis and modeling of impulsive
noise in broad-band powerline communications," in IEEE Transactions on
Electromagnetic Compatibility, vol. 44, no. 1, pp. 249-258, Feb. 2002, doi:
10.1109/15.990732.
[37] Hisashi Sawada, Hiroshi Kuriyamat, Naoki Yusat, Tadanori Mizuno,
and Hiroshi Mineno. “Mutually Complementary Communication Proto-
col based on Destination Oriented Directed Acyclic Graph.” In: (2012), pp.
275–279.
[38] Oana Balmau, Dacfey Dzung, Abdulkadir Karaagac, Vukasin Nesovic,
Aleksandar Paunovic, Yvonne Anne Pignolet, et al., "Evaluation of RPL for
medium voltage power line communication", 2014 IEEE International Con-
ference on Smart Grid Communications (SmartGridComm), pp. 446-451,
2014.
[39] Yvonne-Anne Pignolet, Ilias Rinis, Dacfey Dzung and Abdulkadir Karaa-
gac, "Heterogeneous Multi-interface Routing: Networking Stack and Simu-
lator Extensions", 2012 IEEE 9th International Conference on Mobile Adhoc
and Sensor Systems (MASS), no. Supplement, pp. 16, 2012.
[40] A. N. Abbou, Y. Baddi and A. Hasbi, "Routing over Low Power and
Lossy Networks protocol: Overview and performance evaluation," 2019 In-
ternational Conference of Computer Science and Renewable Energies (ICC-
SRE), Agadir, Morocco, 2019, pp. 1-6, doi: 10.1109/ICCSRE.2019.8807584.
[41] Thubert, "Objective Function Zero for the Routing Protocol for Low-
Power and Lossy Networks (RPL)", RFC 6552 (Proposed Standard), March
2012.
[42] J. Tripathi, J. C. de Oliveira and J. P. Vasseur, "Applicability Study of
RPL with Local Repair in Smart Grid Substation Networks," 2010 First IEEE
International Conference on Smart Grid Communications, Gaithersburg,
MD, 2010, pp. 262-267, doi: 10.1109/SMARTGRID.2010.5622054.
5.2. Future directions 57

[43] Nguyen Thanh Long, Marie-Paule Uwase, Jacques Tiberghien and Kris
Steenhaut, "QoS-aware cross-layer mechanism for multiple instances RPL",
2013 International Conference on Advanced Technologies for Communica-
tions (ATC 2013), pp. 44-49, 2013.
[44] H. Philipps, "Modeling of powerline communication channels", Proc.
3rd Int. Symp. Powerline Communications and Its Applications, pp. 14-21,
1999.
[45] Martocci, J Mil, Pieter Riou, N Vermeylen, W. (2010). Building au-
tomation routing requirements in low-power and lossy networks.
[46] A. S. Kassab , K. G. Seddik, A. Elezabi and A. Soltan, "Realistic Wireless
Smart-Meter Network Optimization Using Composite RPL Metric," 2020
8th International Conference on Smart Grid (icSmartGrid) , Paris, France,
2020, pp. 109-114, doi: 10.1109/icSmartGrid49881.2020.9144938.