EC2V63 - WIRELESS SENSOR

NETWORK DESIGN

UNIT III
6LOWPAN
Contents

6LoWPAN Architecture - protocol stack, Adaptation Layer, Link layers –

Addressing, Routing - Mesh-Under - Route-Over, Header Compression -
Stateless header compression - Context- based header compression,
Fragmentation and Reassembly , Mobility – types, Mobile IPv6, Proxy
Home Agent, Proxy MIPv6, NEMO –Routing – MANET, ROLL, Border
routing.

2
1. 6LoWPAN Architecture
 6LoWPAN (IPv6 over Low-Power Wireless Personal Area Network) is a
networking protocol designed to enable communication between low-power
devices over wireless connections using IPv6. It targets devices that have
limited processing power, memory, and energy resources, making it suitable for
applications like home automation, industrial control, healthcare monitoring,
and sensor networks.

 The Wireless Embedded Internet is created by connecting islands of wireless

embedded devices, each island being a stub network on the Internet.

 A stub network is a network which IP packets are sent from or destined to, but
which doesn’t act as a transit to other networks.
3
4
 The 6LoWPAN architecture is made up of low-power wireless area networks
(LoWPANs), which are IPv6 stub networks.

 A LoWPAN is the collection of 6LoWPAN Nodes which share a common IPv6

address prefix.

 Three different kinds of LoWPANs have been defined: Simple LoWPANs,

Extended LoWPANs, and Ad hoc LoWPANs.

 An Ad hoc LoWPAN is not connected to the Internet, but instead operates

without an infrastructure.

 A Simple LoWPAN is connected through one LoWPAN Edge Router to

another IP network.
5
 A backhaul link (point-to-point, e.g. GPRS) but this could also be a
backbone link (shared).

 An Extended LoWPAN encompasses the LoWPANs of multiple edge routers

along with a backbone link (e.g. Ethernet) interconnecting them.

 LoWPANs are connected to other IP networks through edge routers.

 The edge router plays an important role as it routes traffic in and out.

 A LoWPAN consists of nodes, which may play the role of host or router, along
with one or more edge routers.

 In order to facilitate efficient network operation, nodes register with an edge

router. 6
 These operations are part of Neighbor Discovery (ND), which is an important
basic mechanism of IPv6.

 Neighbor Discovery defines how hosts and routers interact with each other on
the same link.

 LoWPAN Nodes may participate in more than one LoWPAN at the same time
(called multi-homing), and fault tolerance can be achieved between edge
routers.

 A multihop mesh topology within the LoWPAN is achieved either through link-
layer forwarding (called Mesh-Under) or using IP routing (called Route-Over).

 Each LoWPAN Node is identified by a unique IPv6 address, and is capable of

sending and receiving IPv6 packets. 7
 Typically LoWPAN Nodes support ICMPv6 traffic such as “ping”, and use the
user datagram protocol (UDP) as a transport.

 The Simple LoWPAN and Extended LoWPAN Nodes can communicate with
either of the servers through their edge router.

 A LoWPAN Edge Router is typically connected to the Internet over a

backhaul link such as cellular or DSL.

 In an Extended LoWPAN configuration multiple edge routers share a

common backbone link and collaborate by sharing the same IPv6 prefix,
offloading most Neighbor Discovery messaging to the backbone link.

 This greatly simplifies LoWPAN operation as IPv6 addresses are stable

throughout Extended LoWPAN and movement between edge routers is simple.
8
 6LoWPAN does not require an infrastructure to operate, but may also
operate as an Ad hoc LoWPAN.

 In this topology, one router must be configured to act as a simplified edge

router, implementing two basic functionalities: unique local unicast address
(ULA) generation and handling 6LoWPAN Neighbor Discovery registration
functionality.

1. The protocol stack

 The Internet Model is sometimes referred to as a “narrow waist” model, as
the Internet Protocol ties together a wide variety of link-layer technologies with
multiple transport and application protocols.
9
10
 A simple IPv6 protocol stack with 6LoWPAN (also called a 6LoWPAN
protocol stack) is almost identical to a normal IP stack with the following
differences.

 6LoWPAN only supports IPv6, for which a small adaptation layer (called the
LoWPAN adaptation layer) has been defined to optimize IPv6 over IEEE
802.15.4.

 The most common transport protocol used with 6LoWPAN is the user
datagram protocol (UDP), which can also be compressed using the LoWPAN
format.

 The transmission control protocol (TCP) is not commonly used with

6LoWPAN for performance, efficiency and complexity reasons.
11
 The Internet control message protocol v6 (ICMPv6) is used for control
messaging, for example ICMP echo, ICMP destination unreachable and
Neighbor Discovery messages.

 Application protocols are often application specific and in binary format,

although more standard application protocols are becoming available.

12
 Adaptation between full IPv6 and the LoWPAN format is performed by routers
at the edge of 6LoWPAN islands, referred to as edge routers.

 LoWPAN adaptation in an edge router typically is performed as part of the

6LoWPAN network interface driver and is usually transparent to the IPv6
protocol stack itself.

1.2 The Adaptation Layer

 To transport IP packets across the subnetworks, an "IP-over-X" specification is
needed.

 In many cases, more effort is required to map IP layer services to those of the
lower layer. This kind of specification can sometimes become a distinct
(sub)layer, often referred to as an adaptation layer.
13
 An “IP-over-X” adaptation layer needs to map IP datagrams to the services
provided by the subnetwork, which is usually considered to be at layer 2 (L2)
of a layered reference model.

A number of problems may need to be solved here:

 Links in networking can be either point-to-point or multi-access, On multi-

access links, link-layer addressing (L2 addressing) is used.

 Once the IP layer determines the next hop's IP address for a packet, the
adaptation layer's role includes finding the appropriate link-layer address.

 This link-layer address ensures the packet reaches the right destination on its
journey to the intended IP-layer endpoint.
14
 In some cases, the subnetwork might not immediately offer a route for
packets to reach the next IP node.

 For example, when mapping IP to connection-oriented networks like ISDN or

ATM, the adaptation layer could be responsible for establishing and
managing connections as well as deciding when to terminate them.

 Even in non-connection-oriented scenarios like LoWPANs, situations can arise

where the adaptation layer plays a role in determining the next link-layer
hop.

 For instance, in a Mesh-Under configuration, the adaptation layer might need

to identify the next L2 hop and provide it with information about the onward
path for packet forwarding.
15
 The IP packet needs to be packaged (encapsulated) in the subnetwork in such a
way that the subnetwork can transport it and the L2 receiver can extract the IP
packet again.

 Different link layers can transport various packet types beyond IP datagrams,
often indicating next-layer packet type, unlike IEEE 802.15.4 where
6LoWPAN autonomously distinguishes packet encapsulations.

 IP packets might exceed the capacity of layer 2's transport units. To

accommodate larger IPv6 packets, it's necessary to split L3 packets into
multiple L2 packets. The subsequent IP node must then reassemble these
parts to reconstruct the IP packet, known as segmentation and reassembly, or
in 6LoWPAN's terms, fragmentation and reassembly.

16
  IP is designed for standalone packets, leading to potentially redundant
headers. In LoWPAN, a 48-byte IP/UDP header in an IEEE 802.15.4 packet
occupies much payload space, often necessitating fragmentation. A better
solution is 6LoWPAN's approach: reducing redundancy between L3 and
L2 via header compression.

1.2 Link layers for 6LoWPAN

 One of the most important functions of the Internet Protocol is the
interconnection of heterogeneous links into a single interoperable network,
providing a universal “narrow waist”.

 The most basic requirements for a link layer to support 6LoWPAN are
framing, unicast transmission and addressing.
17
 Addressing is required to differentiate between nodes on a link, and to form
IPv6 addresses which are then elided by 6LoWPAN compression.

 It is highly recommended that a link supports unique addresses by default (e.g.

a 64-bit extended unique identifier), to allow for stateless autoconfiguration.

 A link should provide payload sizes at least 30 bytes in length.

 Although UDP and ICMP include a simple 16-bit checksum, it is

recommended that the link layer also provides strong error checking.

 Links include strong encryption and authentication.

18
1.2.1 IEEE 802.15.4

 The IEEE 802.15.4 standard defines low-power wireless embedded radio

communications at 2.4 GHz, 915 MHz and 868 MHz.

 More recently the IEEE 802.15.4a standard was released, extending

802.15.4with two new physical layer options: Chirp Spread Spectrum at 2.4
GHz and Ultra Wide-Band at 3.1–10.6 GHz.

 In practice IEEE 802.15.4 at 2.4 GHz is used almost exclusively today as it

provides reasonable data rates, and can be used globally.

 The 802.15.4 standard provides 20–250 kbit/s data rates depending on the
frequency.
19
20
 Channel sharing is achieved using carrier sense multiple access (CSMA), and
acknowledgments are provided for reliability.

 Link-layer security is provided with 128-bit AES encryption.

 The physical layer payload is up to 127 bytes, with 72–116 bytes of payload
available after link-layer framing, addressing, and optional security.

 The MAC run in two modes: beaconless mode and beacon-enabled mode.

 Beaconless mode uses pure CSMA channel access and operates quite like IEEE
802.11 without channel reservations.

 Beacon-enabled mode uses a hybrid time division multiple access (TDMA)

approach, with the possibility of reserving time-slots for critical data.
21
1.2.2 Sub-GHz ISM band radios

 The sub-GHz ISM bands cover 433 MHz, 868 MHz and 915 MHz and are
especially popular in low-power wireless embedded applications such as
telemetry, metering and remote control.

 The main reasons for sub-GHz popularity are the better penetration of lower
frequency, resulting in better range compared to 2.4 GHz, and the 2.4 GHz ISM
band becoming very crowded in urban environments.

 One example of a popular sub-GHz chip is the Texas Instruments CC1101

transceiver.

 When designing a link layer for this transceiver, the IEEE 802.15.4 frame
structure and beaconless mode operation is typically used as a starting point.
22
1.2.3 Power line communications

 6LoWPAN also has interesting uses over special wired communication links,
such as lowrate power line communications (PLC).

 Applications of this technology include home automation, energy efficiency

monitoring and smart metering.

 Example is watt pulse communication (WPC) technique, greatly reducing the

complexity of communications.

23
1.3 Addressing
 IPv6 addresses are typically formed automatically from the prefix of the
LoWPAN and the link-layer address of the wireless interfaces.

 The difference in a LoWPAN is with the way low-power wireless technologies

support link-layer addressing; a direct mapping between the link-layer address
and the IPv6 address is used for achieving compression.

 Low-power wireless radio links typically make use of flat link-layer

addressing for all devices, and support both unique long addresses (e.g. EUI-
64) and configurable short addresses (usually 8–16 bits in length).

 IPv6 addresses are 128 bits in length, and consist of 64-bit prefix part and a
64-bit interface identifier (IID). 24
 Stateless address autoconfiguration (SAA) is used to form the IPv6 interface
identifier from he link-layer address of the wireless interface.

 For simplicity and compression, 6LoWPAN networks assume that the IID has a
direct mapping to the linklayer address, therefore avoiding the need for address
resolution.

 The IPv6 prefix is acquired through Neighbor Discovery Router

Advertisement (RA) messages as on normal IPv6 link.

 The construction of IPv6 addresses in 6LoWPAN from known prefix

information and known link-layer addresses allows a high header
compression ratio.

25
2. Forwarding and Routing
 Packets will often have to traverse multiple radio hops on their way through the
LoWPAN.

 This involves two related processes: forwarding and routing.

 Both can be performed at layer 2 or at layer 3.

 In each node, the routing protocol fills in a routing information base (RIB),
which contains all the information needed to run the routing protocol.

 The RIB usually an be simplified to a forwarding information base (FIB),

which is consulted when a packet arrives that needs to be forwarded.
26
 Some routing protocols fill the FIB proactively, i.e., the FIB should always
contain an entry for each packet that can actually be forwarded, while others
operate reactively by filling in gaps in the FIB only as packets arrive.

27
 Figure 2.5 shows the usual illustration for routing at the network layer:

 Packets are sent through some link and arrive at a router on an interface if0.

 The router looks up the destination address in its FIB, selects an interface to
forward it out on (here: if1), together with a linklayer address to send it to, and
sends the packet encapsulated with the new link-layer address out via that
interface.

 In LoWPAN, the interface that the packet arrives on at the router node is
usually the same interface that is used again for sending it out has been called
“router on a stick”. It is shown in Figure 2.6

28
29
2.1 L2 forwarding (“Mesh-Under”)
 When routing and forwarding happen at layer 2, they are performed based on
layer-2 addresses, i.e., 64-bit EUI-64 or 16-bit short addresses.

 Internet Engineering Task Force (IETF) is not usually working on layer-2

routing (“mesh routing”) protocols.

 In Figure 2.7 Routing and forwarding is happening at layer 2 essentially

invisible to the LoWPAN adaptation layer.

 In case the actual link-layer forwarding is not hidden from the LoWPAN
adaptation layer (Figure 2.8) ,the link-layer headers describe the source and
destination addresses for the current layer-2 hop.
30
31
32
 To forward the packet to its eventual layer-2 destination, the node needs to
know the final destination address and originator address.

 Since each forwarding step overwrites the link-layer destination address by

the address of the next hop and the link-layer source address by the address of
the node doing the forwarding, this information needs to be stored somewhere
else.

 So, 6LoWPAN defines the mesh header for this.

 In addition to the addresses, the mesh header stores a layer-2 equivalent of an

IPv6 Hop Limit.

33
 Since the diameters of useful wireless multihop networks are usually small, the
format is optimized for hops left values below 15 by only allocating four bits
to that value (first case in Figure 2.9).

 If a value of 15 or larger is needed, the 6LoWPAN packet encoder needs to

insert an extension byte (second case in Figure 2.9).

 The V and F bits indicate whether the originator (or “very first”) address and
the final Destination address, respectively are 16-bit short (1) or 64-bit EUI-
64 (0) addresses.

34
35
2.2 L3 routing (“Route-Over”)
 In the "Route-Over" approach, each routing node in the network makes its
routing decisions based on the Layer 3 (network layer) information. In other
words, every individual router in the path makes a routing decision, thus the
name "Route-Over."

 Layer-3 Route-Over forwarding does not require any special support from the
adaptation layer format.

 Before the layer-3 forwarding engine sees the packet, the adaptation layer has
done its work and decapsulated the packet.

 Fragmentation and reassembly are performed at each hop in Route-Over

forwarding. 36
37
3. Header Compression
 Much of the information that is repeatedly sent in sequences of network
headers is redundant, i.e. could be compressed away.

 Data compression techniques such as gzip and its underlying DEFLATE

algorithm are optimized for eliminating redundancies in a given data item or
stream.

 A separate set of techniques has evolved that focuses on compressing the stacks
of headers in sequences of packets: header compression.

 Most header compression schemes operate hop-by-hop, i.e. as part of the

adaptation layer.
38
 This allows compressing the full header stack including the IP header
immediately before sending the packet on a link, and decompressing and thus
reconstructing the header stack in full before the packet is possibly routed and
sent on via a different link, quite likely with a different header compression
scheme or at least different header compression parameters.

39
 ROHC (robust header compression) standards and its predecessors focus on
compressing flows of packets, such as the sequence of packets from a single
TCP connection or an RTP ( Real-time Transport Protocol) stream.

 This works by setting up flow state for each new connection/stream, and
stepping through a set of compression states during the exchange of the initial
packets of the flow until a high level of compression is attainable.

 The main problem that flow-based header compression needs to solve is that
the per-flow state, the context, is likely to get out of sync between the sender
and the receiver when packets get lost.

 Various techniques such as positive acknowledgments or optimistic

compression with checksums and negative acknowledgments either avoid
progressing the context state in this case or allow quick recovery (robustness).
40
 Instead, the original 6LoWPAN format standard exclusively employs stateless
compression. Without state, there is no synchronization problem, and the
algorithms become very simple.

 Unfortunately, there is no way to compress the large globally routable IPv6

addresses without at least some state.

 So, The 6LoWPANWG has therefore agreed to standardize a second header

compression method, context-based header compression.

1. Stateless header compression

 The 6LoWPAN format specification defines two header compression schemes
that are designed to work together: HC1 to compress IPv6 headers, and HC2
to compress UDP headers.
41
 HC1 is selected by using a dispatch byte of LOWPAN_HC1 (01000010).

 The objective of HC1/HC2 was to enable header compression in an entirely

stateless fashion.

 In other words, there is no requirement for previous agreement between

nodes exchanging the compressed packets.

 HC1 allows the compressor to select elision of each half of both source and
destination address independently by using two bits in the HC1 header each,
called SAE (source address encoding) and DAE (destination address
encoding).

 The rest of the HC1 header is concerned with the compression of the non-
address components of an IPv6 header and illustrated in Fig. 2.12 42
43
 The version number is obviously always 6 and therefore never needs to be
sent.

 Two fields that often are zero in IPv6 headers are the traffic class and the
flow label.

 The C bit in the HC1 header, if set, indicates that these bits indeed are zero
and are not sent. If the C bit is clear, they are included among the non-
compressed fields in the sequence explained below.

 The payload length can be inferred from the remaining length of the
6LoWPAN PDU and therefore is never sent in a compressed header.

 The next header field is one full byte, but has a number of values that are
much more likely than others.
44
  Finally, the Hop Limit was considered to be too difficult to compress and
therefore is always sent in-line in the non- compressed fields.

 The non-compressed fields that follow the HC1 or HC1/HC2 header always
start with the Hop Limit (8 bits). After that it follow

 source address prefix (64 bits), if the high-order bit of SAE is zero;
 source address interface identifier (64 bits), if the low-order bit of SAE is
zero; 45
 destination address prefix (64 bits), if the high-order bit of DAE is zero;
 traffic class (8 bits) and flow label (20 bits), if the C bit is zero;
 next header (8 bits), if NH is zero;
 any non-compressed fields left by HC2, if present;
 any next headers and payloads not further subject to compression.

46
3.2 Context-based header compression
 In many LoWPANs, most packets will go from LoWPAN hosts to some nodes
external to the LoWPAN or in the inverse direction.

 Both cases cannot work with link-local addresses; both source and
destination addresses will be routable addresses.

 Generally, only eight of the 32 bytes of IPv6 addresses in such packets will be
compressible.

 In order to enable the compression of global addresses, this new method

relies on nodes to share some extra information when they connect to the
LoWPAN.
47
 This extra information, called context, is used to make the addresses
compressed when nodes communicate within the LoWPAN.

 The new context-based header compression scheme does not define how the
context is acquired, but the assumption is that this will be done using
6LoWPAN’s extensions to Neighbor Discovery.

 The context stays synchronized between compressor and decompressor.

 To reduce the probability of damage, LoWPAN Nodes should use context-based

header compression only when a higher-layer protocol is in use that protects
the IPv6 addresses using some form of pseudo-header-based checksum and/or
authenticator, such as UDP, TCP or some application-specific integrity
protocol.
48
 The context-based header compression is divided into a compression scheme
for the IP header (LOWPAN_IPHC) and an optional compression scheme for
the next header (LOWPAN_NHC).

 The LOWPAN_IPHC base header needs 13 bits. 49

 The C bit (called CID in the specification controls whether a third byte is added
to specify Context IDs for the source and destination address. If C is not set, the
“default context” is used for both (i.e., the context with context ID = 0).

 The TF bits control how the IPv6 header fields traffic class and flow label are
handled.

50
 The N bit controls both whether a next header field is sent in-line and whether the
next header uses LOWPAN_NHC.

 The compression of the source & destination addresses is controlled by

S/SAM and D/DAM, respectively.

 A special flag, M, can be set to indicate that the destination address is a

multicast address;

 The S flag and the D flag control whether compression of the respective address
is context-based or not.

 Figure 2.16 shows the LOWPAN_NHC base header values for UDP.

51
 The P field and C bit control how
the port number and checksum
fields of the UDP header are
compressed.

 The C bit controls whether the

UDP checksum is elided.

52
If it is set, the UDP checksum is removed by the compressor and must be
recomputed at the decompressor.

The P field controls the compression of the source and destination fields of
the UDP header (see Figure 2.18). 53
54
4. Fragmentation and Reassembly
 IP packets vary in size, with IPv6 packets having a minimum size of 40 bytes,
consisting of a 40-byte IP header and a payload length of 0 bytes.

 IPv6 packets can be exceptionally large, thanks to the jumbogram option,

allowing payloads of up to 4,294,967,295 bytes.

 Most subnetworks, like Ethernet, have a much smaller maximum transmission

unit (MTU) of 1500 bytes, making it challenging to efficiently transport large
IPv6 packets.

 This presentation addresses MTU-related issues in the context of 6LoWPAN,

IPv6, and even IPv4 to illustrate design decisions.
55
 On a host with a single interface, determining the MTU is straightforward,
enabling applications to adjust packet sizes accordingly.

 However, dealing with multiple interfaces or changing MTUs along the route
becomes complex.

 IPv4 accommodates larger packets but requires the ability to fragment and
reassemble them, with a minimum requirement of 68 bytes. 56
 The don't fragment (DF) flag and path MTU detection (PMTUD) are used in
IPv4 to handle MTU variations along the route and prevent fragmentation issues.

 PMTUD has led to a minimal fragmentation rate in typical Internet traffic, with
less than 1% of packets being fragmented.

 IPv6 simplifies fragmentation by eliminating IPv4's fragmentation fields

and relies on a minimum MTU of 1280 bytes, encouraging subnetworks to
support larger packets natively or use IPv6 fragmentation.

57
4.1 The fragmentation format
 6LoWPAN has distinct requirements for fragmentation and reassembly
compared to ATM due to differences in link layer capabilities and the absence
of in-sequence virtual circuit semantics.

 6LoWPAN's link layer provides error checking and variable-length frames,

eliminating the need for an upper-layer equivalent of AAL5.

 Unlike ATM, 6LoWPAN links do not guarantee in-sequence delivery of frames,

necessitating additional information for reassembly.

 Instead of using a "more-fragments" flag, 6LoWPAN copies the size of the

entire packet (IPv6 header + IPv6 payload) into every fragment,
simplifying buffer allocation. 58
 6LoWPAN uses a 16-bit datagram_tag, along with sender and destination link
layer addresses and datagram_size, to distinguish different packets for
reassembly.

 An 8-bit datagram_offset indicates the position of a fragment in the

reassembled IPv6 packet, counting in 8-byte units.

 The first byte of a fragment (dispatch byte) indicates the frame type; some
values are allocated for fragments with datagram_size, reducing the need for
padding.

 Most fragments in 6LoWPAN will be initial fragments with a fragment offset

of zero, allowing for an optimization to elide this number.

59
60
 A procedure is outlined for sending large 6LoWPAN PDUs that are too big to
fit into a link-layer frame, involving calculations for packet size, header size, and
available frame space.

 The process of fragment reception and reassembly involves building a four-

tuple, handling initial and non-initial fragments differently, and ensuring
proper alignment and overlap detection.

4.2 Avoiding the fragmentation performance penalty

 Fragmentation is considered undesirable due to several reasons, including the
decoupling between the unit of loss (fragment) and the unit of
retransmission (entire packet), which can lead to inefficiencies.
61
 In resource-constrained embedded environments, managing resources allocated
to reassembly buffers for fragmented packets is difficult. The uncertainty of
when remaining fragments will be received adds complexity, making reception
unreliable for battery-operated systems with limited RAM.

 Determining the available space in a link-layer frame (max_frame) is

challenging. Additionally, predicting the effectiveness of header compression
further complicates this.

 For applications with larger packet size requirements, techniques like

fragment acknowledgment and retransmission may be necessary for reliable
transmission across multihop mesh paths.

62
5. Mobility
 Mobility in 6LoWPAN refers to the ability of nodes within such networks to
move, change their point of attachment to the network, or even change their
network domain while maintaining their connectivity and communication
capabilities.

 A wireless embedded network in such an application may be used in several

ways, including to track the fork-lift itself, the goods being moved and
personnel at the plant.

 All of these uses of the wireless network require us to deal with node mobility
between edge routers in the same LoWPAN, between LoWPANs and
possibly between network domains.
63
 At the same time, active data flows may be in progress and application servers
may need to know how to reach tracked devices.

1. Mobility types
 Roaming: A process in which a mobile node moves from one network to
another, typically with no existing packet streams.

 Handover: A process in which a mobile node disconnects from its existing

point of attachment and attaches itself to a new point of attachment.

Mobility can alternatively be described with the terms micro-mobility and macro-
mobility.

64
 Micro-mobility refers to mobility that occurs within a network domain. In
6LoWPAN we can consider micro-mobility to refer to the mobility of a node
within a LoWPAN where the IPv6 prefix does not change.

 Macro-mobility on the other hand refers to mobility between networks. In

6LoWPAN we consider macro-mobility to refer to mobility between
LoWPANs, in which the IPv6 prefix changes.

 In wireless networks there are a number of things that may cause a network
to make a change in topology. The causes of topology change can be
categorized simply as

 Physical movement  Sleep schedules

 Radio channel changes  Node failure
 Network performance 65
66
 Node mobility: When a single
node moves between points of
attachment.

 Network mobility: An entire

network moves its point of
attachment.

67
5.2 Mobile IPv6
 The mobility of well-known IP address nodes on the Internet can be dealt
with at the network layer using a protocol called Mobile IP (MIP).

 This new version takes advantage of IPv6 mechanisms and provides route
optimization.

 The goal of MIP is to deal with the mobility roaming problem by allowing a
host to be contacted using a, regardless of its location on the Internet.

 Mobile IP does this using the concept of a home address, which is associated
with a host’s home network.

68
 When a host is away from its home network, and attaches to another
network domain (called the visited network), the new IP address it configured
there is called its care-of address.

 A node communicating with a mobile node roaming in a visited network is

called the correspondent node.

 Normally forwarding in IP networks is handled only by routers, whose route

tables are maintained by routing protocols.

 Mobile IP works using a special kind of routing functionality, which is host

controlled.

 This concept is called a binding, and is implemented by an entity called the

Home Agent (HA). 69
 The Home Agent is responsible for maintaining a binding between a node’s
permanent home address and the temporary care of address used while
roaming in a visited network.

5.2.1 Functionality
 When a mobile node moves away from its home network, it sends a MIPv6
binding update message to its Home Agent (HA) to inform it of the change
in location.

 If the mobile node doesn't know the location of its HA or its home prefix, there
are methods to discover both to ensure proper communication.

 A bidirectional IPv6-in-IPv6 tunnel is established between the HA and the

mobile node for exchanging data packets.
70
71
 Incoming packets addressed to the mobile node's home address in its home
network are intercepted by the HA and encapsulated with the mobile node's care-
of address for delivery.

 To optimize routing and enable direct communication between the mobile node
and corresponding nodes, MIPv6 includes route optimization and security
association procedures. Once established, communication can proceed directly
between these nodes.
The use of MIPv6 as defined in with 6LoWPAN has the following problems:

 IPv6-in-IPv6 tunneling between the HA and LoWPAN Node would incur large
header overheads.

 The traffic burden caused by MIPv6 may be too much for low-bandwidth
wireless links. 72
 The added complexity of implementing MIPv6 in terms of code size and RAM
may be unjustifiable for LoWPAN Nodes.

 Route optimization adds an even greater burden on nodes, as state must be

maintained for every active correspondent node.

5.3 Proxy Home Agent

 A proxy Home Agent (PHA) is an entity which performs MIPv6 functions
on behalf of a local mobile node, interacts with the actual Home Agent of
the node, and handles route optimization on its behalf.

 This greatly simplifies the functions that a mobile node needs to perform to
participate in MIPv6.
73
74
 The PHA is located in the visited network where a mobile node is roaming, in
6LoWPAN this would logically be the LoWPAN Edge Router.

 A PHA acts like a normal MIPv6 host, but additionally performs binding
updates, HA tunneling and route optimization on behalf of other nodes.

 In order to use the PHA concept with 6LoWPAN, a mechanism for registering
with the PHA would need to be defined for use inside the LoWPAN.
(6LoWPAN-ND Node Registration message).

 It include the Home Agent’s address or home prefix, the node’s home address
and some credentials.

75
5.4 Proxy MIPv6
 Proxy MIPv6 (PMIPv6), uses a local hierarchical structure of routers to handle
mobility on behalf of nodes.

 It allows LoWPAN Edge Routers or other local routers to proxy MIPv6 on

behalf of attached LoWPAN Nodes.

 PMIPv6 domain is controlled by a local mobility anchor (LMA).

 The LMA function is usually combined with HA functionality.

 The LMA handles the local mobility of nodes with the help of mobile access
gateways (MAGs), which are points of attachment supporting PMIPv6.
76
77
Mobile access gateways (MAG) send proxy binding updates to the LMA on
behalf of mobile nodes attached to them.

Using bidirectional tunnels built between each MAG and the LMA, the LMA is
then able to forward traffic to the mobile node always using its static address
(known as a mobile node home address).

A binding in the LMA is made between this address and the temporary
address from the visited MAG (the proxy care-of address).

PMIPv6 makes use of RS/RA exchanges directly between the mobile node and
MAG in order to detect when the mobile node has changed its point of
attachment.

78
Limitations

 The RS/RA exchange defined is not compatible with a multihop Route-Over

LoWPAN.

 PMIPv6 is meant to provide a separate 64-bit prefix for each mobile node.

 PMIPv6 only enables a node to talk with its point of attachment (default router).

5.5 NEMO
 Network mobility (NEMO) is a solution for dealing with network mobility
problems, when a router and the nodes attached to it, move their point of
attachment all together.
79
 The philosophy behind NEMO is to extend Mobile IP so that each node does not
need to run Mobile IP, instead only the router they are attached to runs
Mobile IP.

 This philosophy fits the 6LoWPAN model perfectly as LoWPAN Nodes are not
capable of dealing with MIPv6.

 The NEMO protocol introduce a new logical entity called mobile router, which
is responsible for handling MIPv6 functions for the entire mobile network.

 Mobile nodes which are part of the mobile network are called mobile network
nodes (MNNs).

 NEMO extends the functionality of the Home Agent to be able to deal with
prefixes in addition to home addresses of mobile nodes. 80
81
 A mobile router functions like a normal MIPv6 host setting up a bidirectional
tunnel with its Home Agent, but in addition it negotiates prefixes to be
forwarded to it by the Home Agent.

 The Home Agent then forwards all packets matching the bound prefix (therefore
packets for the MNNs) to the mobile router.

 A special flag in the binding update allows the mobile router to indicate it wants
prefix forwarding, and a prefix option lets it configure prefixes with the HA.

Drawback of NEMO
 It can not deal with individual node mobility on behalf of the LoWPAN Nodes.

 NEMO starts to become complicated when mixing different kinds of node

mobility along with PMIPv6. 82
6. Routing
 IP routing deals with maintaining routing tables on IP routers which indicate
which next-hop forwarding decision should be made for the destination of an IP
packet.

 IP routing protocols maintain these routing tables using a wide variety of

techniques ranging from ad hoc dynamic routing for wireless mesh networks to
path vector inter-domain routing on the core Internet.

 There are two main classes of routing protocols useful for 6LoWPAN:
Distance-vector routing and
Link-state routing
83
Distance-vector routing

 Using this approach, each link (and possibly node) is assigned a cost, using
appropriate route metrics.

 When sending a packet from node A to node B, the path with the lowest cost is
chosen.

 The routing table of each router keeps soft-state route entries for the
destinations it knows about, with the associated path cost.

 Routing information is updated either proactively (a priori) or reactively (on-

demand) depending on the routing algorithm.

84
Link-state routing

 In this approach, each node acquires complete information about the entire
network, called a graph.

 To do this, each node floods the network with information about its link
information to nearby destinations.

 After receiving link-state reports from sufficient nodes, each node then
calculates a tree with the shortest-path (least cost) from itself to each
destination using e.g. Dijkstra’s algorithm.

85
6.1 MANET
The mobile ad hoc network (MANET) has three common protocols:

 The ad hoc on-demand distance vector (AODV),

 Dynamic MANET on-demand (DYMO) and
 Optimized link-state routing (OLSR) protocols.

1. Ad hoc on-demand distance vector (AODV)

 The ad hoc on-demand distance vector (AODV) protocol enables mobile ad

hoc multihop networks by quickly establishing and maintaining routes between
nodes, even with quickly changing (dynamic) topologies.
86
 AODV creates routes to destinations when needed for data communications
(reactive), and only maintains actively used routes.

 It includes methods for local repair, and includes a destination sequence number
to ensure loop-free operation.

 AODV is purely a route table management protocol; after routes have been
established they are simply used by IP for forwarding.

 A route request (RREQ) is broadcast throughout the network in order to find

paths to a destination.

 This is responded to with a route reply (RREP) by an intermediate router or by

the destination.
87
 The route error (RERR) message is used to notify about broken links along a
path.

 Figure 4.10 shows an example of reactive route discovery and forwarding in an

ad hoc network.

88
6.1.2 DYMO
 A new reactive distance-vector routing protocol called the dynamic MANET
on-demand (DYMO) protocol has been developed by making improvements
on previous protocols such as AODV and dynamic source routing (DSR).

 This protocol makes use of the same types of route discovery and maintenance
messages as AODV. 89
The main improvements compared to previous work include:

 improved convergence in dynamic topologies

 use of the common MANET packet format
 support for a wide range of traffic flows
 consideration for Internet interconnectivity
 takes hosts and routers into account

6.1.3 OLSR
 The MANET WG has also produced a proactive link-state routing protocol
called the optimized link-state routing (OLSR) algorithm.

 This algorithm applies optimization to the classical link-state algorithm for use
in mobile ad hoc networks
90
 In order to build link-state tables, OLSR routers regularly exchange topology
information with other routers.

 The flooding of this information is controlled by the use of selected

multipoint relay (MPR) nodes.

 OLSR is best suited for relatively static ad hoc networks, thus minimizing the
number of link-state updates throughout the network, which can cause a lot of
overhead.

91
6.2 The ROLL routing protocol
 The routing over low power and lossy (ROLL) networks was formed to
analyze the requirements for and standardize a routing protocol for embedded
applications such as urban ubiquitous networks, industrial automation, building
automation and home automation.

 Low-power and lossy networks (LLNs) are typically made up of embedded

devices with limited processing, memory and power resources.

 The applications and link-layers considered by ROLL have several specialized

properties:

 Traffic patterns are not only peer-to-peer unicast flows, but more often
point-to multipoint or multipoint-to-point flows. 92
  Routers in LLNs have a very small, hard bound on state (limited
memory).
 Most LLNs must be optimized for energy consumption.
 Security and manageability are extremely important as LLNs are typically
autonomous.
 The application spaces aimed at by ROLL are heterogeneous.
 Each may need a different set of features along with routing metrics to
fulfill its requirements.

ROLL architecture

 The architecture of LLNs is very different from that considered by MANET

protocols.

93
 The ROLL protocol can be classified as a proactive distance-vector algorithm
with advanced options for constraint-based routing, multi-topology routing
and traffic engineering.

 The key requirements or assumptions for LLNs that affect the routing
architecture are:
 LLNs are Internet-connected stub networks, with support for multiple
points of attachment.
 Support for dynamic topologies and mobility is required.
 Support is required for multipath routing, and thus multiple forwarding
options.
 Support is required for multiple node and route metrics, and their application
in constraint-based and multi-topology routing.
 Routers in LLNs have limited memory resources.
 Most applications will require enterprise-class security.
94
95
 Routers with interfaces to the LLN and another IP link are called LLN border
routers (LBRs).

 Several LBRs connecting an LLN to a backhaul or backbone link.

 Inside the LLN, the network is made up of LLN routers and LLN hosts.

 Hosts do not participate in the LLN routing algorithm, and instead simply
choose default LLN routers.

 Routing protocol uses a graph structure between nodes and LBR.

 After the basic topology is constructed, the routing protocol maintains

upstream (from node to LBR) and downstream (from LBR to node) paths.
96
 There are two concepts important to understanding ROLL protocol operation:

Metric granularity:
 ROLL uses the concept of a very granular route metric called depth.
 This metric is used by the basic ROLL protocol mechanisms for building the
graph, making use of siblings and for loop avoidance.

Routing time scale:

 ROLL makes routing decisions on two different time scales: route setup time
and packet-forwarding time.
 In route-setup time the routing protocol maintains the basic graph topology and
routing tables using static or slowly moving metrics, which is a continuous
process.
 ROLL enables packet-forwarding time decisions to be made using dynamic
metrics on a packet-by-packet basis. 97
Building and maintaining topology

 Low-power and Lossy Networks (LLNs) initiate the creation of a routing

topology by forming a dynamic graph directed toward exit points (LBRs)
using IPv6 Neighbor Discovery messages.

 The topology automatically constructs multiple distance-vector paths from

nodes to LBRs, facilitating node-to-LBR packet forwarding and allowing for
dynamic node repositioning and topology merging in ad hoc environments.

 The ROLL topology is efficiently maintained using periodic advertisements,

avoiding overhead by leveraging existing ND signaling. Node dissemination
of path cost information ensures efficient downstream routing, contributing to a
comprehensive global understanding for multi-topology routing and traffic
engineering. 98
Forwarding traffic

99
 The ROLL protocol maintains a fundamental topology enabling upstream
forwarding from LLN nodes to Local Border Routers (LBRs).

 Reliability is improved by offering multiple routing paths to nodes, achieved

by selecting from multiple default routers at packet-forwarding time, preventing
application process failure.

 Downstream forwarding to LLN nodes from LBRs involves various methods

such as hop-by-hop distance-vector state, source routing, or a combination.

 The decision is made at packet-forwarding time, considering metrics,

constraints, and the available memory in routers.

 Source routing can be stateless or utilize states and flow labeling for path
indication.
100
Optimizations

Traffic engineering:
 To meet specific requirements like traffic engineering, solutions in ROLL may
include off-line traffic engineering for quality of service, accomplished by
labeling, multi-topology routing, or source routes to designated nodes.

Node-to-node flows:
 ROLL addresses diverse needs, such as node-to-node flows, with optimizations
like boulevard routes or reactive distance-vector features.

Mobility support:
 Additionally, mobility support (MIPv6, NEMO, MANEMO) integration with
ROLL is explored to enhance adaptability for various applications.
101
6.3 Border routing
 The ROLL protocol is designed to optimize traffic flows in Low-power and
Lossy Networks (LoWPANs), especially for applications with traffic coming
from or going towards the Internet.

 The border router, typically an edge router in 6LoWPAN, faces challenges in

maintaining route entries between its interfaces belonging to different routing
domains, particularly when dealing with different routing protocols.

 Border routing, common on the broader Internet, becomes relevant at the edge,
especially with the emergence of IP routing in wireless stub networks using
mesh routing protocols.

102
Three cases of border routing in
6LoWPAN are considered:

 Simple LoWPAN,
 Extended LoWPAN, and
 Route redistribution

103
 In Simple LoWPANs, with a single edge router, prefix-based route entries are
used between the LoWPAN and IPv6 link.

 Filtering mandates the use of a routing protocol only over specific access
networks.

 In Extended LoWPANs, routing between the LoWPAN and IP network may

occur in different places, requiring destination route entries.

 Route redistribution is considered for optimal routing algorithms between

interfaces, such as OLSR and open shortest path first (OSPF).

104