Received May 17, 2018, accepted June 7, 2018, date of publication July 3, 2018, date of current version August

20, 2018.
Digital Object Identifier 10.1109/ACCESS.2018.2847765

Route Optimization of Electric Vehicles

Based on Dynamic Wireless Charging
DIMITRIOS KOSMANOS1 , LEANDROS A. MAGLARAS2 , (Senior Member, IEEE),
MICHALIS MAVROVOUNIOTIS 3 , SOTIRIS MOSCHOYIANNIS4 ,
ANTONIOS ARGYRIOU1 , (Senior Member, IEEE),
ATHANASIOS MAGLARAS5 , (Member, IEEE), AND HELGE JANICKE 2
1 Department of Electrical and Computer Engineering, University of Thessaly, 382 21 Volos, Greece
2 Department of Computing Technology, De Montfort University, Leicester LE1 9BH, U.K.
3 KIOS Research and Innovation Center of Excellence, Department of Electrical and Computer Engineering, University of Cyprus, 1678 Nicosia, Cyprus
4 Department of Computer Science, University of Surrey, Guildford GU2 7XH, U.K.
5 Department of Electrical Engineering, Technological Educational Institute of Thessaly, 411 10 Larissa, Greece

Corresponding author: Helge Janicke ([email protected])

This work was supported by the U.S. Department of Commerce under Grant BS123456.

ABSTRACT One of the barriers for the adoption of electric vehicles (EVs) is the anxiety around the limited
driving range. Recent proposals have explored charging EVs on the move, using dynamic wireless charging
which enables power exchange between the vehicle and the grid while the vehicle is moving. In this paper,
we focus on the intelligent routing of EVs in need of charging so that they can make most efficient use of
the so-called mobile energy disseminators (MEDs) which operate as mobile charging stations. We present a
method for routing EVs around MEDs on the road network, which is based on constraint logic programming
and optimization using a graph-based shortest path algorithm. The proposed method exploits inter-vehicle
communications in order to eco-route electric vehicles. We argue that combining modern communications
between vehicles and state of the art technologies on energy transfer, the driving range of EVs can be extended
without the need for larger batteries or overtly costly infrastructure. We present extensive simulations in
city conditions that show the driving range and consequently the overall travel time of electric vehicles is
improved with intelligent routing in the presence of MEDs.

INDEX TERMS Constraint solving, dynamic wireless charging, electric vehicles, inductive power transfer,
optimization, vehicular communications routing.

I. INTRODUCTION can facilitate data access, which is fundamental to realizing

There is an increasing interest among government agencies, the premise of smart cities.
research institutions and industry around the globe in improv- Smart vehicles are expected to be a part of a Vehicular
ing urban living while reducing the environmental impact. Ad hoc NETwork (VANET), a mobile ad hoc network of
The term smart city has been coined to describe the city of cars that has been proposed to enhance traffic safety and
tomorrow in which modern intelligent technologies, such as provide comfort applications to drivers. A VANET has some
IT communication systems, sensors, machine learning, data unique characteristics such as high mobility of nodes, while
analytics, come together to provide better services to the cars must follow predefined routes; messages that come
citizens. Just like a complex system, a smart city can monitor, from several applications, with different priority levels; high
coordinate and manage information, connectivity and assets interference, in a noisy environment, and so on. Using the
that citizens need every day and adapt to accommodate their on-board unit, vehicles can communicate with each other as
demands. One of the basic components of this environment is well as with road side units (RSUs) enabling smart appli-
envisaged to be the next generation of vehicles that combine cation solutions but also enhanced road safety and traffic
new sensing, communication and social capabilities. By pro- management. According to several works, e.g., see [1], smart
viding mobile wireless sensing and communications, vehicles vehicles exhibit five features: self-driving, safety driving,

VOLUME 6, 2018 This work is licensed under a Creative Commons Attribution 3.0 License. For more information, see http://creativecommons.org/licenses/by/3.0/ 42551
 D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

social driving, electric vehicles and mobile applications.

In this paper, we focus on electric vehicles.
One of the prohibiting factors for the adoption of
the Electric Vehicles (EVs) across Europe is the driving
range [2], [3]. That is, the range the vehicle can cover before
it needs to be recharged. The lack of supporting charging
infrastructure is a pivotal prohibiting factor. The deployment
of charging infrastructure is a hard problem [4] as it inad-
vertently requires changes to the existing civil infrastructure
and these are costly and take a long time to implement. The
car industry is experimenting with larger and more powerful FIGURE 1. Wireless charging of EV using spiral coils.

batteries - new Tesla and Volkswagen (VW) EVs have been

released with powerful batteries that promise to cover up to
400km without intermediate charge. However, it is argued driveway and the increase of the burden on the road traffic
that in the future batteries of reduced capacity should be used, conditions. Buses (inner city) repeatedly move at prescribed
mainly for environmental reasons. routes that are scheduled well in advance. Hence, the EV can
It arises the need for new approaches to charging electric meet them by appointment at specific locations. The process
vehicles that overcome the lack of supporting infrastructure is similar to charging of aircraft in flight. When the bus
and the difficulty of adapting the existing civil infrastruc- finishes its round trip or its energy inventories are depleted,
ture, i.e., road network, without requiring new batteries that it will return to the fixed static charging station where it will
take up most space in the car and are not environmentally either fully charge or change the batteries.
friendly. Dynamic wireless charging is a technology that is In this paper, the focus is on describing the mechanics
still in the R&D phase. A number of companies are actively of the proposed dynamic wireless charging of EVs and the
developing dynamic wireless charging solutions, both in the challenges that arise. An EV in need of charge would typi-
research and testing phases. BMW has already demonstrated cally have a choice of MEDs to which it could attach itself.
wireless charging with the i8 model. Tesla motors also has The main contribution of this paper concerns the intelli-
already produced the Plugless Model S that can use wireless gent routing of EVs in need of charging using a MED or
inductive charging at home. Wireless charging can be the a SCS, and more specifically a solution that draws upon con-
key enabler for electric vehicles if they are to surpass the straint logic programming (CLP) and a graph-based shortest
convenience of gas cars [5]. Preliminary analysis, e.g., see [6], path algorithm (cf Section IV). The optimization problem
suggests that even the most far-out ideas around wire- of (re)routing is considered under a range of criteria and
less charging may become reality sooner than most expect. priorities. The objective of the problem is to route a set of
Qualcom in [7] introduces Wireless Electric Vehicle Charg- EVs in the best possible way, with optimization criterion the
ing (WEVC), which is a simple, no fuss solution for charging waiting time for static charging from a SCS or for dynamic
Electric Vehicles (EVs). Qualcomm Halo WEVC technology charging from a MED with target the minimum overall travel
uses resonant magnetic induction to transfer energy between time, as is depicted in the flow chart of Fig. 2. This procedure
a ground-based pad and a charging pad on the electric vehicle. is described in detail in Section V-B.
The WEVC proposes expensive charging pad that need to be Extensive simulations were conducted in city conditions
installed at the surface of a road track. On the other hand, our in order to evaluate the proposed ‘‘on the move’’ charg-
proposed dynamic charging model uses ordinary city buses ing technique (cf Section V). With different initial energy
as energy sources on the move. conditions for all the EVs of the simulation, two different
This drives the investigation towards integrated solutions charging systems are compared: one uses a static charging
that allow EVs to charge on the move. In [8] and [9], station (SCS) only and the second combines a SCS with
Maglaras et al. have proposed a novel idea for increasing a MED. The experiments show that the driving range and
the driving range without requiring a significant change consequently the overall travel time is improved by about four
in existing road infrastructure. The idea builds on deploy- times in the dynamic charging system involving MEDs.
ing buses and heavy goods vehicles (HGVS), large goods This paper proposes a dynamic wireless charging method
vehicles (LGVs) or trucks, as mobile charging stations, the for EVs using the city buses as MEDs. Our proposed tech-
so-called Mobile Energy Disseminators (MEDs) [8]. While nique is the first in literature according to our knowledge that
a bus is moving along its normal route an EV in need of uses the ordinary city buses and trucks for recharging needs
charging attaches itself to it and charges via wireless power of EVs without any extra infrastructure. This paper makes the
transmission, as shown in Fig. 1. following contributions:
The buses that take the role of MEDs are ordinary city • A dynamic wireless method is proposed for the
buses that follow their predefined routes at the roads of the charging of the EVs on the move. This method uses
city. We do not use extra buses or trucks as dynamic energy existing city buses as MEDs, without the need for under-
disseminators that could have as a result the occupancy of one ground or other infrastructure;

42552 VOLUME 6, 2018

D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

FIGURE 2. Overall flow chart for the MED or SCS selection and travel time minimization.

• Our proposed dynamic wireless charging method is in motion, the overall travel time can be improved
based on wireless V2V communications and uses a route about four times compared with the use of only static
optimization solution. The usage of the wireless com- stations.
munication among EVs and MED coordinates the real- • Our intelligent route search method takes into account
time booking procedure for either the SCS or the MED, the waiting time either for the MED - EV appoint-
optimizing the waiting time; ment or for the waiting time at the queue of SCS.
• CACC technology for the MED - EV synchronization We show that our method decreases significantly the
during the time that EV follows the MED is used; waiting time for the charging procedure and the charging
• In addition, we show that vehicles can extend their travel time that is needed for an EV, because the EV is charging
range using a real-time Energy exchange that can be when it continues its route;
facilitated by a (IPT) process.; • The improvement of travel time and driving range of
• Combining static charging stations with Mobile energy electric vehicles comes with a negligible cost in travel
disseminators that can be used for dynamic charging distance. Starving vehicles do not have to stop or make

VOLUME 6, 2018 42553

 D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

long re-routes to find a stationary station and recharge and transportation network traffic conditions. In [19] a route
their batteries; search method for electric vehicles (EVs), which calculates
The remainder of the paper is organized as follows. the minimum travel time that includes necessary stops to
Section II discusses related work and places the research static charging stations, is proposed. The above method uses
within that of the wider community. Section III introduces only static charging stations in which the electric vehicles
the key concepts and the overall architecture of the proposed must stop for charging. On the other hand, we provide the
system. Section IV presents the problem formulation of rout- possibility of dynamic wireless charging of an electric vehicle
ing electronic vehicles given the presence of static and mobile on the move, following a MED for a part of its predefined
stations. Section V presents simulation parameters, describes route at the roads of the city. Our method introduces for
the evaluation of the method and discusses economic benefits the first time in the literature the concept of Mobile Energy
of the proposed method. Section VI concludes the article. Disseminators that can take the role of energy sources and can
operate along with static charging stations in order to decrease
II. RELATED WORK the overall waiting time before the charging procedure begins
The wireless power transmission technology is being applied and the charging time that is needed for an EV. Based on the
for a number of years now in many areas of electrical appli- work in [17] dynamic wireless charging of vehicles promises
ances, like speakers, music and sound transmission gener- to partially or completely eliminate the overnight charging
ally, alarm systems, electric bells, and electrical facilities of electric vehicles through the use of dynamic chargers that
of low power in general. In the field of wireless charg- may be installed on the roads to keep the vehicle batteries
ing of electric vehicles, there are many architectures and continuously charged, thus making electric vehicles more
special experimental systems that have already been pro- attractive. The use of dynamic wireless charging may increase
posed, built and implemented (e.g. Korea reports [10]). driving range and reduce the size of the battery pack of an
In some of these infrastructures the locations (points) used electric vehicle. On the other hand, this leads to increased
for charging are either fixed (static stations) installed either safety concerns and infrastructure costs.
under the surface of streets and in other public locations Previous work on dynamic wireless charging has not con-
(i.e. garages) or on lightning columns [11], [12] on the road sidered the solution of moving energy charging stations that
side. Specifically, regarding the underground electric coils can charge vehicles, which are also on the move, in order to
installed under the surface of streets, KAIST proposes a new reduce the range anxiety and increase the reliability of EVs.
design concept for an alternate electric car-On-Line Elec- Mario et al. [11] presented a solution called Telewatt that
tric Vehicle (OLEV) [13]. OLEV draws its electric power involves the reuse of existing public lighting infrastructure
from underground electric coils without using any mechan- for vehicle charging. It does so by exploiting the excessive
ical contact. But, a great concern using these approaches is power of the lamps mostly at night. This system that sup-
the electromagnetic field exposed to the people that move ports wireless charging between the infrastructure and the
around these streets. OLEV has also a small battery, which moving vehicles raises health issues related to the leaking
enables the vehicle to travel on roads without the underground magnetic flux. In [20] Ning et al. present a system that
electric coil. Batteries are recharged whenever OLEV draws can charge vehicles through inductive coupling. In [21] a
electric power from the underground coils and thus, do not non-radiative energy transformer that can perform efficient
require expensive separate charging stations. However, above wireless energy transfer, commonly referred as Witricity and
technology can be effective if about 30% of the roads in based on ‘‘strong coupling’’ between two coils which are
Seoul have the underground electric power coil, which is separated physically by medium-range distances is investi-
quite costly. The wireless power transmission in the proposed gated. The prototype for EV that was developed at Oak Ridge
system is achieved using the Tesla coil method, with spiral National Laboratory (ORNL) in the United States achieved
coils installed on the vehicles. efficiency of nearly 90% for 3 kW power delivery. However,
Previous work on charging electric vehicles mainly focuses systems that are based on inductive coupling between the grid
on static charging stations [14], swappable batteries [15], eco- and a moving car can cause power pulsations in the vehicle
routing of vehicles [16] or dynamic charging [17] that is based battery and the grid supply. This can result in deterioration on
on static sources. In [18] a routing strategy for vehicle charg- the battery service life of EVs as well as a drop on the power
ing called ‘‘Charging Station Strategy - Vehicle Powertrain quality of the grid [22].
Connected Routing Optimization (CSS-VPCRO)’’ is pro- The disadvantages of these methods can be summarized as
posed. This approach constitutes solving an iterative least cost follows.
vehicle routing process, which utilizes the communication of • Charging an EV from a stationery charger introduces a
electrified vehicles (EVs) with competing charging stations large or small delay due to
to exchange data, such as electricity price, energy demand, 1) the change of the route of the movement of the EV
and time of arrival. EV routing problem is solved to minimize to the loading point (location),
the total cost of travel using the Dijkstra algorithm with the 2) the need of parking for a sufficient period of time
input from EVs battery management system, electricity price to charge, and
from charging stations, powertrain component efficiencies, 3) the restoration of the EV at the initial route.

42554 VOLUME 6, 2018

D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

• The infrastructure would need to be extensive and con- TABLE 1. Miles per 10-minute charge for electric cars [29]
*This is for a 30 amp public charging station.
sequently expensive [23]
• The (energy transfer efficiency) performance of the
charging method would be relatively insufficient
(or low) due to the inherent operational difficulties of
the systems (e.g., distance, parallelism, etc.)
The solution we propose in this paper builds on the use of
inner city buses as MEDs, hence it does not suffer from the
pitfalls associated with static charging stations. In addition, public wired charging station to provide just 22 miles of range
it uses buses or trucks for the dynamic charging, so predefined to an electric car.
moving charging stations which have predefined scheduled
routes along the existing road network, rather than vehicle- B. EVS AND MEDS IN A VANET ARCHITECTURE
to-vehicle (V2V) charging schemes that have been discussed The use of mobile nodes as relay nodes is common in
in the literature [24]. vehicular ad hoc networks (VANETs). In a VANET, mobile
The EVs attach themselves to one or more MEDs during nodes can serve as carriers or disseminators of useful
some part of their journey and until they have enough energy information [30]. Defining influential spreaders, nodes that
to reach their destination (or get to the closest static charging can disseminate the information to a large part of the net-
station). In this way, electric cars are charged ‘‘on the fly’’ work effectively, is an open issue in ad hoc networks [31].
and their range is increased while moving along the road. In VANETs, nodes with predefined or repeating routes that
Hence, our proposal does not require significant changes to can cover a wide range of a city region can play the role of
the existing road network and civil infrastructure [11], [25], roadside units in terms of message dissemination. By exploit-
[26] and, unlike other proposals [27], does not pose any health ing their mobility these disseminating nodes can provide even
hazards. higher quality-of-service (QoS).
Following a similar approach the proposed dynamic wire-
III. DYNAMIC CHARGING AND MOBILE ENERGY less charging system is using special nodes, buses or trucks,
DISSEMINATORS that act as energy sources to EVs that are in energy need.
The dynamic wireless charging system is based on the com- The architecture of the proposed system is shown in Fig. 3.
bination of vehicular communications and inductive power These vehicles, which are called MEDs, use electric plug
transfer (IPT) among the energy carriers and the electric in connection or IPT in order to refill starving EVs. Buses
vehicles. IPT allows efficient and real-time energy exchange can play the role of MEDs in urban environments, since they
where the vehicles involved can play an active role in the follow predefined scheduled routes and their paths cover a
procedure. major part of a city, while trucks can play the role of energy
chargers mainly on highways. Buses can be fully charged
A. ENERGY TRANSFER VIA IPT when parked, before beginning their scheduled trip, and can
Using the IPT wireless method, a 10-minute charge would be continuously charged along their journey by IPT stations
provide a driver with an energy charging of 3 - 8 kWh of installed at bus stops (See Fig. 3).
electric energy, which is equivalent to about 9 - 23 miles EVs follow the MED for a part of its route in order to
travel distance. The United States fuel economy esti- perform dynamic wireless charging. Specifically, vehicles
mates that 35 kWh energy charging equals with 100 miles follow the MED with the same speed while charging, using
travel distance. The energy charging 3 - 8 kWh requires the Cooperative Adaptive Cruise Control (CACC) technol-
20 - 50 kW charging rate from the moving charging sta- ogy [32]. CACC is an enhancement of Adaptive cruise con-
tions (see Table 1). This travel distance corresponds to trol (ACC), which is based on sensor data. It leads to tighter
30 - 78 percent of the drivers average daily travel distance. following gaps between EV - MED and faster response to
In real-world terms, that means typical urban American velocity changes compared to ACC, and makes collabo-
drivers could cover 78 percent of their average daily travel rative driving such as platooning feasible [33]. Using the
of 23 miles on a 10-minute charge with charging rate 50 kW. CACC technology, vehicles that book charging places on the
European drivers fare even better; a 10-minute charge with same MED can create clusters/platoons where the MED will
charging rate 50 kW under this wireless scenario would play the role of the clusterhead [34]. The wireless commu-
cover nearly two days of a typical European’s driving habits, nication that is needed for motion synchronization between
which amounts to about 20 kilometers or 12.5 miles per EV and MED is carried out with beacons that are periodic
day [28]. single-hop messages.
In the case that the charging rate would be 20 kW, The buses or trucks (MEDs) run on electric power. They
a 10-minute charge would cover about 9 miles or about will have battery systems for their movement, which are
15 kilometers. By comparison, a public 30 amp wired charg- used exclusively by the bus or truck (MED). At the same
ing station provides electric cars with just 3.7 miles of range time they carry other systems of special batteries with more
on a 10-minute charge; it takes about an hour at a typical energy, which will only be used for charging of EV vehicles

VOLUME 6, 2018 42555

 D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

4) Vehicles come close and using wireless communica-

tion, like an advanced cruise control system, controlled
for safety reasons by the MED and while in motion,
come in such a position that the smaller diameter’s
MED coil comes close enough to or enters in the
EV coil.
In this article we investigate the wireless power trans-
fer via a loose connection device consisting of 2 coils,
of plain or conical form. Due to the design of the coils,
and the very close position between them, the energy trans-
fer efficiency will be more than 90%. The proposed design
has a similar functionality with the typical Tesla coils [35],
or Rogowski Coil [36]. With this solution, electric power is
transferred to the vehicle through an electrically generated
magnetic field. The basic functionality of the charging pro-
cess is comparable to charging via a cable. An innovative,
induction-based mechanism that is developed by Siemens
eCar Powertrain Systems [37] can be also used to offer a
significantly higher degree of convenience, when compared
with charging via a cable.
A major concern when dealing with strong magnetic fields,
such as those used in wireless power transfer, has to do
with the impact on living organisms. By only turning on the
coils when a compatible electric vehicle is over the primary
charging pad, the charging system eliminates the possibility
FIGURE 3. Application example of a Mobile Energy Disseminator: In A, that a person or animal could be affected by the strong fields
Contactless Wireless-Consistency charging is used to deliver charging
to a bus; in B, V2V communication between MED and EV; in C, EV created. Another issue with safety has to do with the presence
recharges from the bus using IPT. of metal objects at or close to primary charging pads. These
objects can cause hazardous conditions and can interfere
in motion. The energy of these batteries will be able to with WPT. To address this problem, a foreign object detection
cover the energy need of several EVs. The total energy of system can be deployed in future to determine when objects
the charging batteries of the bus is expected to be greater are on top of the primary coils. In such situations the system
than 200 kWh. The energy of the batteries of an ordinary will not energize the transmitting coil so as to avoid damage
EV is about 50 kWh, hence the amount of the energy of to the vehicle and/or charging system.
the batteries carried by the bus will be capable to serve
4 EVs for a total recharge and more for partial recharges. C. COMMUNICATION AMONG ENTITIES
The charging rate will be 20-50 kW (cf. Section 1), so that To state its presence each MED or SCS periodically broad-
the required charging voltage shall be relatively low. Finally, casts cooperative awareness messages (CAM). Each beacon
the bus or truck will carry the mechanisms necessary for the message consists of a node identifier (Vid), node location,
connection and transfer of energy from the MED to the EVs. scheduled trip (a subset of set L), current charging capa-
The EV charging process will be as follows. bility (CC) and energy value (E=kWh), the queue time at
1) EV contacts MED and makes an appointment (time, SCS or waiting time (wt) at MED appointment point and the
location). speed value of the MED. CC is the current energy that
2) EV drives near the MED and creates a platoon with it the mobile charging station can afford to dispose of to charge
to initiate the charging process. the vehicle without jeopardising its own needs. These mes-
3) The MED charges the EV via a loose connection sages are disseminated by all vehicles that effectively act as
device consisting of 2 coils, of plain form or better relay nodes.
of spiral conical form for greater efficiency and ease
of connection. These coils can be of different diam- IV. ROUTING EVS IN NEED OF CHARGING
eter and perhaps even of different numbers of turns. A. PROBLEM FORMULATION: CONSTRAINED
Another solution comprises using 2 coils, one of which SHORTEST PATH
with larger diameter which is mounted on the EV The problem of routing EVs can be presented using a directed
and the other with the smaller diameter on the MED. weighted graph. Let G = (N , A) be a weighted graph where
E / M Shielding will be available on all 2 vehicles. The N is a set of points, e.g., road intersections or Static Charging
2 coils will be properly covered, and will be uncovered Stations (SCS) and A = {(i, j) | i, j ∈ N , i 6= j} is a set
during the charging process. of arcs (links) connecting two points. SCSs are defined as

42556 VOLUME 6, 2018

D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

TABLE 2. Mathematical symbols used in this paper.

S = {s0 , . . . , sv } and a set of dummy nodes that represent at different SCS or MED points (if needed). The energy level
possible multiple visits to the same static recharging station at point i is defined as ik . Hence, the initial energy level is
is defined as S 0 = {sm+1 , . . . , sm+h } such that S ∪ S 0 ⊆ N . defined as sk .
Each SCS i is associated with a waiting time wti . Let xijk and ykij be binary decision variables that define
An EV can also receive energy by MEDs that visit a whether EV k passed from point i to j and whether EV k
predefined cyclic route of MED points M = {m0 , . . . , mu }. received energy from a MED from point i to j, respectively.
Similarly with the SCS, a set of dummy nodes may represent Also, let zki and qki be binary decision variables that defines
possible multiple visit to the same MED point defined as M 0 the SCS where EV k received energy and the MED point
such that M ∪ M 0 ⊆ N . An EV can attach to a MED at any where EV k attached with a MED, respectively. All variables
point in its route and start charging. Note that the charging rate used in this paper are summarised in Table 2.
of MED is always higher than the consumption rate. Similar The objective to minimize the travel time of EVs is given
with the SCS, each MED point i has a waiting time wti This next:
is because an EV may need to wait to a point until a MED is X X X
available or arrives. MEDs and SCSs accept/reject demands min ( (dtij xijk ) + (cti + wti )zki
of EVs in an intelligent way, i.e., to minimize the route of k∈K (i,j)∈A,i 6 =j i∈S∪S 0
X
the vehicles at the best possible way or to distribute energy + (wti qki )) (1)
at the best possible way (defined by the communication X
i∈M ∪M 0
X
system). s.t. xijk − xjik
Each arc (i, j) ∈ A is associated with a non-negative j∈N j∈N
travel time dtij ∈ R+ and a non-negative energy needed to

k
1, if i = s ;
travel cij ∈ R+ when points i and j are connected otherwise

= −1, if i = t k ; ∀i ∈ N , ∀k ∈ K (2)
dtij = cij = ∞. The weight matrix of the problem is defined 
0, otherwise,

as D = {dtij }n×n .
The objective of the problem is to route a K set of EVs in xij − ykij ≥ 0, ∀k ∈ K , ∀j ∈ N , ∀i ∈ N , i 6= j,
k
(3)
the best possible way, i.e., minimum travel time. The problem jk ≤ ik − (cij )xijk + (ρ2 dij )ykij xijk + Qk (1 − xijk ),
can be formulated as a multiple constrained shortest path
∀k ∈ K , ∀j ∈ N , ∀i ∈ N , i 6= j, (4)
problem. Every k-th EV has a battery of Qk capacity, starting
point sk and destination point ek . The travel time is defined ik ≥ 0, ∀k ∈ K , ∀i ∈ N , (5)
by the driving (dt), the charging (ct) and waiting times (wt) ik ≤ Qk , ∀k ∈ K , ∀i ∈ N , (6)

VOLUME 6, 2018 42557

 D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

ik = Qk zki , ∀k ∈ K , ∀i ∈ S ∪ S 0 , (7) Algorithm 1 FindShortestPath(k,sk ,ek )

ik ≥ cij , ∀k ∈ K , 1: INPUT EV information, e.g., id, source and destination
2: FinalRoutek ← ∅ % final route of k EV
∀i ∈ N , ∃j ∈ S ∪ S 0 ∪ M ∪ M 0 , i 6 = j, (8)
3: T k ← 0 % travel time of k EV
xijk , ykij ∈ {0, 1}, ∀k ∈ K , ∀i ∈ N , ∀j ∈ N , i 6 = j, (9) 4: Rk ← Dijkstra(sk , ek ) % partial route of k EV
zki ∈ {0, 1}, ∀k ∈ K , ∀i ∈ S ∪ S 0 , (10) 5: if (Rk is feasible) then
qki ∈ {0, 1}, ∀k ∈ K , ∀i ∈ M ∪ M , 0
(11) 6: FinalRoutek ← Rk
7: else
where cti is the charging time from a charging station or visit i 8: p ← FindBestEnergyPoint(sk )
(for a MED the charging time is already embedded to the 9: FinalRoutek ← Dijkstra(sk , p) ∪ Dijkstra(p, ek )
tour in (4), and wti is the waiting time at charging station 10: end if
(or a MED’s point) i. 11: T k ← Cost(FinalRoutek )
Constraint (2) ensures flow conservation of the route; con- 12: OUTPUT FinalRoutek % feasible route to travel verified
straint (3) ensures that whenever an EV receives energy from by (12)
a MED while moving always consumes energy; constraint (4) 13: OUTPUT T k % travel time using (2) but for a single EV
ensures that an EV has enough energy to move to the next
point (including MED’s points); constraint (5) and (6) ensures
that energy level never falls under zero or exceeds its capac-
ity; constraint (7) ensures that an EV is fully charged at satisfies (12), meaning that it has sufficient energy to reach
static energy station; constraint (8) ensures that an EV has the destination. If the route is feasible then the EV should
enough energy to reach at least one recharging static station or begin its route without any energy recharging consideration.
MED point. Otherwise, it needs to find a point, either static or moving, to
The feasibility of an EV k route can be identified by the recharge its battery in order to have sufficient energy to reach
current energy level and the total energy needed for the route the destination as shown in Algorithm 1.
such that energy must not be negative, as follows: For this case Dijkstra’s algorithm is used again to find
  the best point to receive energy from. Since there may
X be several static charging stations or MED points that the
sk −  cij  + ρ ≥ 0 (12) EV can choose, several Dijkstra’s calculations are performed,
(i,j)∈A one for each point, and the best one is selected as shown
where sk
is the initial energy level, cij the energy consumed in Algorithm 2. The criteria to identify the best energy
from points i to j and ρ is the induced energy. recharging point depends on the total travel time, including
The key differences of the proposed shortest path problem waiting time, charging time and driving time. In addition,
(described above) with the traditional shortest path problem the energy point selected needs to be feasible, i.e., the current
are: energy level of the EV needs to be enough to reach the
a) multiple shortest paths are required, and selected energy point (i.e., constraint (8)). Hence, the energy
b) energy constraints are imposed points that cannot be reached according to (12) are discarded.
The proposed problem is more challenging and realistic Finally, when the energy point is selected the shortest path
because not all shortest routes are feasible due to the energy using the Dijkstra’s algorithm is calculated from the selected
constraints; see (12) and also one shortest route may affect the energy point to the destination. Note that in case this path
remaining shortest routes. For example, if an EV is currently is still not feasible because the energy level may not be
charging at a SCS; then the other EVs will possibly have to sufficient to travel from the energy point to the destination
wait (i.e., increasing the queue time of the SCS) or find a the process in Algorithm 2 can be repeated from the current
shorter route via another SCS. position, e.g., the energy point to the next energy point.

B. SOLUTION METHOD V. EVALUATION

Since the problem is a shortest path problem it can be solved To evaluate the effect of the dynamic wireless charging
by several existing optimization algorithms efficiently (i.e., of EVs, we conducted simulations in the city of Erlangen.
in polynomial time). In this paper, we consider the well-
known Dijkstra’s algorithm [38] to calculate the shortest A. EVALUATION SETUP
route, e.g., minimize the travel time in Eq 2, for EV k from As can be seen in Fig. 4 a bus which follows a specific route
its starting point sk to its destination point ek . However, (shown in yellow in the Fig.) is used as a MED. On the
the problem has several constraints that need to be addressed other hand, a static charging station (SCS) is located at a
and by simply using the Dijkstra’s algorithm from sk to ek fixed point at the road side of the corresponding city district.
may result to an infeasible route, i.e., (12) does not hold. All the parametric side roads of the area in which the SCS
The key idea of the proposed solution method is to initially and MED charging models are located are used as starting
check whether the route calculated by Dijkstra’s algorithm points (sk ) for the dynamic wireless charging system with the

42558 VOLUME 6, 2018

D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

Algorithm 2 FindBestEnergyPoint(sk ) TABLE 3. Evaluation parameters.

1: INPUT current point of EV k

2: best ← ∅ % route of a best energy point
3: p % best energy point
4: for (i ∈ S ∪ M ) do
5: Rk ← Dijkstra(sk , i)
6: if ((Cost(Rk ) < Cost(best)) && (Rk is feasible)) then
7: p←i
8: best ← Rk
9: end if
10: end for
11: OUTPUT p % best energy point

B. IMPLEMENTATION OF THE DYNAMIC CHARGING

SYSTEM
As described in Section III-C, all the EVs are informed for the
waiting time (wti ) either by the SCS i or by the MED i through
the periodical communication with MEDs or SCS (using the
CAM messages). As an example, assume that an EV k is
located at point (sk ) as starting point in Fig. 4. In order this EV
to decide the best point for the insertion of dynamic charging
system the Dijkstra’s algorithm is used (i.e. the Algorithm 1).
The point (mb ) is the best point for the MED system, while
the point(sb ) is the best point for insertion for the SCS system
(see Fig. 4).
The value for minimization with our dynamic charging
algorithm is the travel time for a vehicle between the starting
point (sk ) and the target point (ek ). The total travel time if
an EV chooses the SCS choice depends on the travel time
FIGURE 4. Part of the Erlangen city map used for conducting the between the (sk ,sb ) points, the charging time at the SCS,
simulations. The MED route is marked in yellow. The position of
the SCS is marked in green. The brown arrows are pointing at the the waiting time here and the travel time between the points
starting points of the journeys of the EVs for both the SCS and (sb ,ek ), for which the Dijkstra’s algorithm is used again.
the SCS + MED charging system.
The charging rate level of the EVs at the SCS is about
19, 2 KW /sec [39]. The waiting time at the SCS depends on
same probability. The point at which the EVs are introduced the queue of the SCS and the driving time between (sk ,sb )
in SCS or MED system is shown in Fig. 4 with (mb , sb points. Each EV periodically informed by the SCS about the
respectively). The number of EVs that are inserted in the current queue and all the bookings that SCS already has with
system is between 0 and 100. In addition, each EV k entering the Queue (Waiting) time (wtb ) variable. Based on its current
the system has starting energy sk according to a uniform distance to SCS and mean velocity it can compute the time
distribution with values between 1 − 6 kWh. that it will arrive to the SCS (Driving time (dt)). So it can
The only communication paths available are via the compute the waiting time as: WaitingTime = wtb −dt(sk , sb ).
ad-hoc network and there is no other communication infras- If a vehicle chooses the MED for its recharging needs,
tructure. All the above parameters and the selected evaluated travel time will be adjusted to reflect the travel time between
area, as well, were not in favour of any charging method the points (sk ,mb ), the waiting time of a vehicle at point mb ,
(MED or SCS). The power of the antenna is Ptx = 18dBm the time interval for which this vehicle follows the MED
and the communication frequency f is 5.9 Ghz. In our sim- and thus is charged (EV k follows the MED for the roads
ulations, we use a minimum sensitivity (Pth) of −69 dBm (i,j) which are defined from the binary variable ykij ) and the
to −85 dB, which gives a transmission range of 130 to travel time from the last point (mj ) of the last road (i,j), where
300 meters, as can be seen in Table 3. As a result of the EV k follows the MED, to the destination point (ek ) with the
above transmission range, there is no communication with a usage of Dijkstra’s algorithm. At the starting point (sk ), each
few EVs. So, a number of EVs are excluded from the charging EV at short intervals informed by the MED about its current
procedure because of the communication lost among EVs. position (mi ) and the booked road segments that MED already
This happens when the Signal-to-Interference-Ratio (SINR) has.
threshold is below 10 dB due to attenuation that is caused by An electric vehicle also computes the closest point (mb ) to
the building obstacles of the city. meet the MED based on the MED’s cycle and the vehicle’s

VOLUME 6, 2018 42559

 D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

current position and the driving time, using mean velocity. relation to the total road segment length and the availability
Based on the charging coefficient a vehicle computes for how of the MED. In order to represent the induced energy per hour
many road segments it needs to follow the MED and that to the EV, (15) is rewritten:
way it can find the ending point (mj ). Based on the book-
cij = P ∗ Tij − ρ (16)
ing of the MED, its current position and meeting point mb ,
a vehicle computes the waiting time (the time that it will In Equation (16) the ρ is the induced energy to the vehicle k
need to wait for the MED to come free of any booking and is given by:
at meeting point (mb )). If road segments (mb , mj ) are not
ρ = tcont ∗ Cind ∗ Pind (17)
booked then the waiting time will be: wt = dtMED(mi , mb )−
dtVehicle(sk , mb ). Cind is the induction coefficient and tcont the time of contact
If the above equation is negative then EV k will have to between the MED and the EV. Pind is the power of the MED.
go for the next cycle of the MED: wt = dtMED(mi , mb ) + The values of the above parameters can be seen in Table 1.
dtMED(mb , mb ) − dtVehicle(sk , mb ). If any road segments We ignore acceleration and deceleration phenomena.
between (mb , mj ) are booked then the specific EV will have
to go for the next cycle of the MED again. We must add that C. STARTING ENERGY VS POWER CONSUMPTION LEVELS
there is no upper limit on the waiting time of a vehicle until the In our simulations, we used 3 levels of recharging needs
MED will be available. All the above procedure of the travel for the sum of the EVs. The starting energy for each EV is
time optimization of an Ev is deeply described in Fig. 2. the remaining energy with which they approach the starting
For the charging time of an EV from the MED, when this points of the system. We consider 3 different levels of the
EV books the MED then it knows the point (mi ), so it can power consumption energy for the EVs in comparison with
compute the charging time based on mean velocity and the their initial energy. At the first level of the recharging energy
ending point (mj ) of charging. In order to calculate the energy we consider that only the 20% of EVs need recharging in
will be needed for each vehicle, the power consumption for oder to reach at their destination (see Fig. 5a). The second
each road traveled must be computed [40]. The energy cost level of the power needs of the EVs is that in Fig. 5c.
of every road segment can be expressed as a proportion of the Here 60% of EVs need recharging, increasing the complexity
mean velocity. The velocity is the quotient of the distance of of the system. Last, at the third level of power need and
the road segment and the time that a vehicle will need to spend initial energy comparison almost all the EVs need recharging
on this segment (i, j), i.e. Ti,j , on average. The two forces (the 95% of EVs), as can be seen in Fig. 5e. Contrary to the [3]
that oppose the motion of an automobile are rolling friction, in which the number of drivers with range anxiety is a fixed
Froll and air resistance, Fair ( [9]). number, this number is dynamic in our system and depending
1 on the EVs needs. All the drivers with initial energy smaller
Froll = µς ∗ m ∗ g, Fair = A ∗ C ∗ p ∗ u2 (13) than the energy will be needed to be consumed are defined as
2
anxious drivers.
where, m is the mass of the car in Kg, g = 9.8m/s2 ,
u is the mean velocity in m/s and µς is the rolling resistance D. SCS VS. SCS + MED
coefficient. C is a dimensionless constant called the drag In this section we conduct a comparison of two different
coefficient that depends on the shape of the moving body, A is charging systems using three scenarios (see
the silhouette area of the car (m2 ) and p is the density of the Fig. 5a,5c and 5e. The first charging system contains only
air (about 1.2 kg/m3 at sea level at ordinary temperatures). a static charging station, and the second charging system has
Typical values of C for cars range from 0.35 to 0.50. In a SCS and a MED. In Fig. 5b,5d,5f the travel time results
constant-speed driving on a level road, the sum of Froll and of the above two system are presented that correspond to the
Fair must be just balanced by the forward force supplied charging needs of Fig. 5a,5c,5e, respectively.
by the drive wheels. The power that a vehicle needs when Studying these results, it transpires that as the charging
traveling with a steady speed is given by (14). needs of vehicles are increasing, the travel time for both
P = n ∗ FForward ∗ u = n(Froll + Fair ) ∗ u (14) systems is also increasing. Specifically at the Level 1 of
recharging needs, the travel time using the dynamic charging
where, n is the efficiency factor of the system. The energy model (SCS + 1 MED) is smaller at about 2 times than that
cost of vehicle k for traveling in road segment (i, j) in kwh, using the charging system (SCS) (see Fig. 5b). At the Level 2
i.e. cij , is calculated by (15). of recharging needs, the corresponding travel time using the
(SCS + 1 MED) model is improved and is now at about
cij = P ∗ Tij (15)
3 times smaller than that using the (SCS) model (see Fig. 5d).
If the road segment belongs to the path of a MED, then a Last at the Level 3 of recharging needs, the travel time using
vehicle can increase its energy by induction. The amount of the dynamic combined charging model is at about 4 times
the induced energy is proportional to the total time that an smaller than that using only the (SCS) model (see Fig. 5f).
EV and the MED will stay connected. This time depends on Another observation from the results is that the travel time
the meeting point (mb ) between a vehicle and the MED in of the (SCS+MED) system is less than the (SCS) for all the

42560 VOLUME 6, 2018

D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

FIGURE 5. Travel Time of all the Levels of Energy recharging. (a) (LEVEL 1): The 20% of EVs need recharging. (b) Travel Time when the 20% of EVs
need recharging. (c) (LEVEL 2): The 60% of EVs need recharging. (d) Travel Time when the 60% of EVs need recharging. (e) (LEVEL 3): The 95% of
EVs need recharging. (f) Travel Time when the 95% of EVs need recharging.

circumstances of anxious drivers (0-100) and energy charg- meet and follow the MED is compared with the queue time
ing need levels. For a small number of anxious drivers the for each EV at the SCS. Moreover, in Fig. 7 the percentage
difference between the two charging systems is very small. of EVs that select the MED or the SCS for recharging is
As the number of anxious drivers is increasing, the difference presented. We can see that as the number of anxious drivers is
between the two systems is increasing too. However, when increasing, the number of EVs that select the MED as energy
the number of anxious drivers is above 50 (for the Level 3 of disseminator is increasing too.
charging needs) the difference is diminished. This behavior is Studying more carefully Fig. 6, it is obvious that at the
due to the waiting time of the vehicles for the MED for a large starting time of the dynamic charging system when the queue
number of cycles because of the preceding MED’s bookings. of the SCS is empty and due to the fact that all EVs select
Last, it is obvious that when the number of anxious drivers the MED for recharging results on the increase of the waiting
is above average of overall EVs the need of a MED in addition time. As the simulation time increases, the waiting time for
to a SCS is necessary, because the difference between the MED and the queue time for SCS both rises and falls irregu-
(SCS+MED) system and the system that has only one SCS is larly. This happens because the choice of EVs (MED or SCS)
bigger with (60%, 95%) anxious drivers than that with (20%). for recharging are quickly interchanged. Studying the travel
time of Fig. (5a,5c,5f), a reduction of the difference of the
E. SCS + MED SYSTEM EVALUATION travel time between the systems (SCS),(SCS + 1 MED)
In this subsection the evaluation of the system (SCS + has observed. This phenomenon can be explained due to
1 MED) is presented in more detail. In Fig. 6, the waiting the increase of the waiting time at the MED, because of
time (wt) of each EV at the point (mb ) that is planned to the frequent MED selection, (see Fig. 6) when the number

VOLUME 6, 2018 42561

 D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

FIGURE 7. The number of EVs that select either the MED or the SCS for
their recharging needs for all the levels of Energy recharging.

F. COST BENEFIT ANALYSIS

There are several revenue possibilities stemming from this
concept as well. Electric utilities, for example, might consider
subsidizing the modification of trucks and buses into MEDs
under a scenario in which the utility then becomes a revenue
sharing partner with the MED owner. Governments at the
state, local and national levels are all involved in policy-
making decisions regarding environmental impact mitiga-
tion options, often using analytical tools [41]. In this case,
governments may consider offering tax incentives to modify
trucks and buses into MEDs to further promote popularity and
adoption of EVs.
In addition there are entrepreneurial advantages. Special
software will need to be designed and refined for the physical
platooning of the MEDs and EVs as well as for handling
the appointment and billing logistics. Manufacturers will be
asked to design and build the magnetic subsystems that create
the foundation of the wireless charging systems. Converting a
bus or truck into a MED would cost around 26, 000, while the
FIGURE 6. Waiting Time for MED vs Queue Time of SCS. (a) Waiting Time cost of adding the technology to a passenger vehicle would be
for MED vs Queue Time of SCS with the (LEVEL 1) of Energy recharging.
(b) Waiting Time for MED vs Queue Time of SCS with the (LEVEL 2) of about 1, 500.
Energy recharging. (c) Waiting Time for MED vs Queue Time of SCS with
the (LEVEL 3) of Energy recharging.
VI. CONCLUSIONS
We have proposed a solution for increasing the driving
range of electric vehicles based on modern communications
of anxious drivers increases (i.e. above 80 anxious drivers between vehicles and state of the art technologies on energy
for Level 3 energy recharging). This leads anxious drivers transfer. The proposed solution steers away from larger and
to choose the SCS and when its queue time increases, this more powerful batteries, although these would still be useful
situation reverses again. and complements what we are proposing here. It does not
Comparing the waiting time results for the 3 Levels of require changes to existing road infrastructure which are
energy recharging, we can see that when the MED takes costly and often pose health hazards. In contrast to vehicle-to-
part more in EVs recharging (see Fig. 6c), the waiting vehicle (V2V) charging schemes that are recently discussed
time or queue time is not increased with such a steep mode in the literature [24], our work builds on the idea of using
as that of Fig. (6a, 6b). Moreover, the more interchanges the city buses that follow predefined scheduled routes for
between waiting time for MED or queue time for SCS dynamic charging in urban environments.
in Fig. 6c and the more usage of MED for EVs charg- Combining modern communications between vehicles and
ing needs at the Level 3 (see Fig. 7) fully justify the state of the art technologies on energy transfer, we have
wide difference between the two charging systems (SCS), shown that vehicles can extend their travel range. Energy
(SCS +1 MED) in Fig. 5f. So the more balanced usage of exchange between vehicles can be facilitated by a process
(SCS), (SCS +1 MED) charging systems, when the recharg- called ‘‘Inductive power transfer’’ (IPT). This allows for an
ing needs of EVs are increased, approves that a combined efficient and real-time energy exchange where vehicles can
charging system such as (SCS +1 MED) is essential. play an active role in the process.

42562 VOLUME 6, 2018

D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

Making use of inductive charging MEDs that act as mobile REFERENCES

charging stations can improve the overall travel time of a fleet
[1] A. M. Vegni, M. Biagi, and R. Cusani, ‘‘Smart vehicles, technolo-
of vehicles compared to using only static charging stations. gies and main applications in vehicular ad hoc networks,’’ in Vehicu-
Specifically, using a MED in support of a SCS the overall lar Technologies—Deployment and Applications. Rijeka, Croatia: InTech,
travel time can be improved about four times compared with 2013.
[2] C. Thiel, J. Krause, and P. Dilara, ‘‘Electric vehicles in the EU from
the only SCS usage case. The improvement of travel time
2010 to 2014—Is full scale commercialisation near,’’ Publications Office,
comes with a negligible cost in travel distance, but starving Luxembourg, Tech. Rep., 2015, doi: 10.2790/311494.
vehicles otherwise would have to stop for a relatively long [3] E. Bulut and M. C. Kisacikoglu, ‘‘Mitigating range anxiety via vehicle-
time or make longer re-routes to find a stationary station and to-vehicle social charging system,’’ in Proc. IEEE Veh. Technol. Conf.,
Jun. 2017, pp. 1–5.
recharge their batteries. [4] A. Hecker and R. Wies, ‘‘Charging infrastructure for EVs in Beijing:
Summarizing, the main findings of the article are: A spatial analysis from real customer data at two districts,’’ in Proc. Int.
Conf. Connected Vehicles Expo (ICCVE), Oct. 2015, pp. 336–341.
1) The concept of MEDs (Mobile Energy Disseminators) [5] G. A. Covic and J. T. Boys, ‘‘Modern trends in inductive power transfer for
is introduced and the performance of such a system transportation applications,’’ IEEE J. Emerg. Sel. Topics Power Electron.,
vol. 1, no. 1, pp. 28–41, Mar. 2013.
is evaluated. MEDs are ordinary buses of a city that
[6] M. Yamauchi. (2017). [Online]. Available: https://www.pluglesspower.
follow predefined scheduled routes, play the role of a com/learn/mainstream-electric-cars-are-headed-towards-wireless-
new kind of dynamic charging stations in urban envi- charging/
ronments without requiring changes to existing road [7] D. Graeme. (2017). [Online]. Available: https://www.qualcomm.com/
news/onq/2017/05/18/wireless-dynamic-ev-charging-evolution-
infrastructure which are quite costly. qualcomm-halo
2) Routing of vehicles is conducted on the basis of opti- [8] L. A. Maglaras, F. V. Topalis, and A. L. Maglaras, ‘‘Cooperative
mizing their total travel time, with the use of an intelli- approaches for dymanic wireless charging of electric vehicles in a smart
city,’’ in Proc. IEEE Int. Energy Conf. (ENERGYCON), May 2014,
gent routing algorithm and through the use of wireless
pp. 1365–1369.
communication between vehicles (V2V). This routing [9] L. A. Maglaras, J. Jiang, A. Maglaras, F. V. Topalis, and S. Moschoyiannis,
procedure takes into account the waiting time for a ‘‘Dynamic wireless charging of electric vehicles on the move with mobile
MED or the queue time in a SCS and overall traffic con- energy disseminators,’’ Int. J. Adv. Comput. Sci. Appl., vol. 30, no. 30,
pp. 239–251, 2015.
ditions of the city with main objective the minimization [10] G. Jung et al., ‘‘High efficient inductive power supply and pickup system
of the travel time for every EV. for on-line electric bus,’’ in Proc. IEEE Int. Electr. Vehicle Conf. (IEVC),
3) The combination of a static charging station (SCS) with Mar. 2012, pp. 1–5.
[11] M. A. Ruiz, F. A. Abdallah, M. Gagnaire, and Y. Lascaux, ‘‘TeleWatt:
a MED improves the overall travel time about four An innovative electric vehicle charging infrastructure over public lighting
times compared to the simple SCS usage case. The systems,’’ in Proc. 2nd Int. Conf. Connected Vehicles Expo (ICCVE),
waiting time and the charging time of every EV are also Las Vegas, CA, USA, Dec. 2013, pp. 741–746.
improved. [12] B. Lane, ‘‘Innovative on street EV charging solutions,’’ Ecolane
Consultancy Next Green Car, Wales, U.K., White Paper, 2015. [Online].
4) The improvement of travel time and driving range of Available: http://www.ecolane.co.uk/wp-content/uploads/2015/01/
electric vehicles comes with a negligible cost in travel Ecolane-Innovative-on-street-EV-charging-solutions.pdf
distance. [13] S. Lee, J. Huh, C. Park, N.-S. Choi, G.-H. Cho, and C.-T. Rim, ‘‘On-line
electric vehicle using inductive power transfer system,’’ in Proc. Energy
As part of our future work, we intend to evaluate our pro- Convers. Congr. Expo. (ECCE), Sep. 2010, pp. 1598–1601.
[14] H. Xu, S. Miao, C. Zhang, and D. Shi, ‘‘Optimal placement of charging
posed dynamic charging method using different evaluation
infrastructures for large-scale integration of pure electric vehicles into
parameters. Specifically, we plan to use a larger number of grid,’’ Int. J. Elect. Power Energy Syst., vol. 53, pp. 159–165, Dec. 2013.
MEDs in combination with existing SCSs and place them [15] J. D. Adler and P. B. Mirchandani, ‘‘Online routing and battery reservations
in different areas of the city in order to further evaluate our for electric vehicles with swappable batteries,’’ Transp. Res. B, Methodol.,
vol. 70, pp. 285–302, Dec. 2014.
dynamic charging system. We also intend to combine the [16] M. M. de Weerdt, S. Stein, E. H. Gerding, V. Robu, and N. R. Jennings,
proposed dynamic wireless charging with a battery swapping ‘‘Intention-aware routing of electric vehicles,’’ IEEE Trans. Intell. Transp.
system, either for the MED buses or for the EVs, along Syst., vol. 17, no. 5, pp. 1472–1482, May 2016.
[17] S. Lukic and Z. Pantic, ‘‘Cutting the cord: Static and dynamic inductive
with a dynamic inventory of fully charged batteries (FBs) for wireless charging of electric vehicles,’’ IEEE Electrific. Mag., vol. 1, no. 1,
recharging electric vehicles depleted batteries (DBs) with the pp. 57–64, Sep. 2013.
minimum charging cost [42] too. [18] M. H. Amini and O. Karabasoglu, ‘‘Optimal operation of interdependent
Moreover, in our future plans is the inclusion of the power systems and electrified transportation networks,’’ Energies, vol. 11,
no. 1, p. 196, 2018.
dynamic electric vehicle charging (DEVC) technology of [19] Y. Kobayashi, N. Kiyama, H. Aoshima, and M. Kashiyama, ‘‘A route
Qualcom along with SCS and MEDs. DEVC technology search method for electric vehicles in consideration of range and locations
allows vehicles to charge while driving. For example, for only of charging stations,’’ in Proc. Intell. Vehicles Symp. (IV), Jun. 2011,
pp. 920–925.
a small part of the city road network ground-based pad can be [20] P. Ning, J. M. Miller, O. C. Onar, C. P. White, and L. D. Marlino, ‘‘A com-
installed in order to allow the technology that uses resonant pact wireless charging system development,’’ in Proc. 28th Annu. IEEE
magnetic induction to transfer energy wirelessly to a pad Appl. Power Electron. Conf. Expo. (APEC), Mar. 2013, pp. 3045–3050.
integrated in the vehicle. That way a more dynamic charging [21] S. L. Ho, J. Wang, W. N. Fu, and M. Sun, ‘‘A comparative study
between novel witricity and traditional inductive magnetic coupling in
system will be designed taking into account the additional wireless charging,’’ IEEE Trans. Magn., vol. 47, no. 5, pp. 1522–1525,
infrastructure cost. May 2011.

VOLUME 6, 2018 42563

 D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

[22] J. M. Miller et al., ‘‘Demonstrating dynamic wireless charging of an elec- DIMITRIOS KOSMANOS received the M.Sc.
tric vehicle: The benefit of electrochemical capacitor smoothing,’’ IEEE degree from the University of Thessaly, Greece,
Power Electron. Mag., vol. 1, no. 1, pp. 12–24, Mar. 2014. where he is currently pursuing the Ph.D. degree
[23] G. R. Nagendra, L. Chen, G. A. Covic, and J. T. Boys, ‘‘Detection of EVs with the Department of Electrical and Computer
on IPT highways,’’ IEEE J. Emerg. Sel. Topics Power Electron., vol. 2, Engineering. Before his current position, he was
no. 3, pp. 584–597, Sep. 2014. collaborating as a Fellow Researcher with the
[24] M. C. Kisacikoglu, A. Bedir, B. Ozpineci, and L. M. Tolbert, ‘‘PHEV-EV CERTH Department of Volos. He was interested in
charger technology assessment with an emphasis on V2G operation,’’ Oak optimized video transmission techniques in wire-
Ridge Nat. Lab., Oak Ridge, TN, USA, Tech. Rep. ORNL/TM-2010/221, less cooperative networks. His current research
Mar. 2012. interests include techniques about improving the
[25] N. Machiels, N. Leemput, F. Geth, J. Van Roy, J. Büscher, and J. Driesen, quality of wireless communication in vehicular ad hoc networks (Vanets)and
‘‘Design criteria for electric vehicle fast charge infrastructure based on the improved detection and suppressing jamming threats in 802.11p WI-FI
flemish mobility behavior,’’ IEEE Trans. Smart Grid, vol. 5, no. 1,
protocol for Vanets and specifically for autonomous vehicles.
pp. 320–327, Jan. 2014.
[26] G. A. Putrus, P. Suwanapingkarl, D. Johnston, E. C. Bentley, and
M. Narayana, ‘‘Impact of electric vehicles on power distribution net-
LEANDROS A. MAGLARAS (SM’15) received
works,’’ in Proc. Vehicle Power Propuls. Conf. (VPPC), Sep. 2009,
the B.Sc. (M.Sc. equivalent) degree in electrical
pp. 827–831.
and computer engineering from the Aristotle Uni-
[27] G. Jung et al., ‘‘Wireless charging system for on-line electric bus(OLEB)
versity of Thessaloniki, Greece, in 1998, the M.Sc.
with series-connected road-embedded segment,’’ in Proc. 12th Int. Conf.
Environ. Elect. Eng. (EEEIC), May 2013, pp. 485–488. degree in industrial production and management,
[28] G. Pasaoglu et al., ‘‘Driving and parking patterns of European car
and the M.Sc. and Ph.D. degrees in electrical and
drivers—A mobility survey,’’ Eur. Commission, Luxembourg, Tech. Rep., computer engineering from the University of Thes-
2012, doi: 10.2790/70746. saly, in 2004, 2008, and 2014, respectively, and the
[29] L. Maglaras. (2017). How to Charge Your Electric Car ‘on the Fly’. Ph.D. degree in intrusion detection in SCADA sys-
[Online]. Available: http://www.brinknews.com/how-to-charge-your- tems from the University of Huddersfield in 2018.
electric-car-on-the-fly/ He is currently a part-time Senior Lecturer with the School of Computer
[30] O. Rehman, M. Ould-Khaoua, and H. Bourdoucen, ‘‘An adaptive relay Science and Informatics, De Montfort University, conducting research in the
nodes selection scheme for multi-hop broadcast in VANETs,’’ Comput. Cyber Security Centre. He served on the Editorial Board of several Interna-
Commun., vol. 87, pp. 76–90, Aug. 2016. tional peer-reviewed journals, such as the IEEE ACCESS and the Journal on
[31] P. Basaras, D. Katsaros, and L. Tassiulas, ‘‘Detecting influential spreaders Security and Communication Networks (Wiley).
in complex, dynamic networks,’’ Computer, vol. 46, no. 4, pp. 24–29,
Apr. 2013.
[32] V. Milanés, S. E. Shladover, J. Spring, C. Nowakowski, H. Kawazoe, MICHALIS MAVROVOUNIOTIS received the
and M. Nakamura, ‘‘Cooperative adaptive cruise control in real traffic B.Sc. degree in computer science from the Uni-
situations,’’ IEEE Trans. Intell. Transp. Syst., vol. 15, no. 1, pp. 296–305, versity of Leicester, Leicester, U.K., in 2008,
Feb. 2014. the M.Sc. degree in natural computation from
[33] A. Tiganasu, C. Lazar, and C. F. Caruntu, ‘‘Design and simulation evalu- the University of Birmingham, Birmingham, U.K.,
ation of cooperative adaptive cruise control for a platoon of vehicles,’’ in in 2009, and the Ph.D. degree in computer science
Proc. 20th Int. Conf. Syst. Theory, Control Comput. (ICSTCC), Oct. 2016, from the University of Leicester in 2013. He is
pp. 669–674. currently with the KIOS Research and Innova-
[34] S. Santini, A. Salvi, A. S. Valente, A. Pescapé, M. Segata, and R. Lo Cigno, tion Center of Excellence, University of Cyprus,
‘‘A consensus-based approach for platooning with intervehicular com- Cyprus. His research interests include evolution-
munications and its validation in realistic scenarios,’’ IEEE Trans. Veh. ary computation, swarm intelligence, memetic computing, combinatorial
Technol., vol. 66, no. 3, pp. 1985–1999, Mar. 2017. optimization problems, heuristics and metaheuristics, artificial neural net-
[35] J. C. Stark, ‘‘Wireless power transmission utilizing a phased array of tesla works, computational intelligence in dynamic and uncertain environments,
coils,’’ Ph.D. dissertation, Massachusetts Inst. Technol., Cambridge, MA, and relevant real-world applications. He is the Vice-Chair of the Task Force
USA, 2004.
on Evolutionary Computation in Dynamic and Uncertain Environments,
[36] J. D. Ramboz, ‘‘Machinable Rogowski coil, design, and calibra- under the Evolutionary Computation Technical Committee of the IEEE
tion,’’ IEEE Trans. Instrum. Meas., vol. 45, no. 2, pp. 511–515,
Computational Intelligence Society.
Apr. 1996.
[37] I. Haq, R. Monfared, R. Harrison, L. Lee, and A. West, ‘‘A new
vision for the automation systems engineering for automotive powertrain
SOTIRIS MOSCHOYIANNIS received the B.Sc.
assembly,’’ Int. J. Comput. Integr. Manuf., vol. 23, no. 4, pp. 308–324,
degree in mathematics from the University of
2010.
Patras, Greece, the M.Sc. degree in information
[38] E. W. Dijkstra, ‘‘A note on two problems in connexion with graphs,’’
systems, and the Ph.D. degree in theoretical com-
Numer. Math., vol. 1, no. 1, pp. 269–271, Dec. 1959.
puter science from the University of Surrey, U.K.
[39] M. Yilmaz and P. T. Krein, ‘‘Review of battery charger topologies,
charging power levels, and infrastructure for plug-in electric and hybrid
He is currently a Senior Lecturer in complex sys-
vehicles,’’ IEEE Trans. Power Electron., vol. 28, no. 5, pp. 2151–2169, tems with the University of Surrey, an Associate
May 2013. Member of the Surrey Centre for Cyber Secu-
[40] X. Wu, D. Freese, A. Cabrera, and W. A. Kitch, ‘‘Electric vehicles’ energy rity, a GCHQ—recognized U.K. Academic Centre
consumption measurement and estimation,’’ Transp. Res. D, Transport of Excellence in Cyber Security Research, and a
Environ., vol. 34, pp. 52–67, Jan. 2015. member of the Nature-Inspired Computer Engineering Research Group.
[41] D. Lloyd, S. Moschoyiannis, N. Elia, A. Penn, and C. Knight, ‘‘A web- He has published over 40 peer-reviewed publications, including three book
based tool for identifying strategic intervention points in complex sys- chapters. His research interests are in the modeling and analysis of complex
tems,’’ in Proc. Games Synthesis Complex Syst. (CASSTING ETAPS), systems and his work on interpretable AI and machine learning as applied to
vol. 220, 2016, pp. 39–52. path-dependency analysis is relevant to the work presented in this paper. He
[42] X. Tan, G. Qu, B. Sun, N. Li, and D. H. K. Tsang, ‘‘Optimal has been involved in a number of research projects and has received funding
scheduling of battery charging station serving electric vehicles based from the EU, U.K. research councils, government departments, and industry.
on battery swapping,’’ IEEE Trans. Smart Grid, to be published, doi: He received five best paper awards at international journals and conferences.
10.1109/TSG.2017.2764484.

42564 VOLUME 6, 2018

D. Kosmanos et al.: Route Optimization of EVs Based on Dynamic Wireless Charging

ANTONIOS ARGYRIOU received the Diploma HELGE JANICKE received the Ph.D. degree in
degree in electrical and computer engineering computer science in 2007. He was involved in
from the Democritus University of Thrace, Xanthi, cyber security with organizations, such as Airbus
Greece, in 2001, and the M.S. and Ph.D. degrees in Group, QinetiQ, Ministry of Defence, and General
electrical and computer engineering (as a Fulbright Dynamics U.K., among others. He is currently
Scholar) from the Georgia Institute of Technology, the Director of the Cyber Technology Institute,
Atlanta, GA, USA, in 2003 and 2005, respectively. De Montfort University (DMU). He is also the
From 2004 to 2005, he was a Senior Engineer with Head of the School of Computer Science and
Soft Networks, Atlanta, GA, USA. From 2007 to Informatics. His interests are covering formal ver-
2010, he was a Senior Research Scientist with ification techniques and their application to cyber
Philips Research, Eindhoven, The Netherlands. He is currently an Assistant security, SCADA and industrial control system security, and aspects of cyber
Professor with the Department of Electrical and Computer Engineering, warfare. He established the DMU’s Airbus Group Centre of Excellence in
University of Thessaly, Volos, Greece. His research interests include wireless SCADA cyber security and forensics research in 2013. He is the General
communication systems and networks, and video delivery. He currently Chair of the International Symposium on SCADA and Industrial Control
serves on the Editorial Board of the Journal of Communications. He has also Systems Cyber Security Research. He is the Editor-in-Chief of the Journal of
served as the Guest Editor for the IEEE TRANSACTIONS ON MULTIMEDIA Special Security and Safety (EAI) and serves on the editorial board and as a reviewer
Issue on Quality-Driven Cross-Layer Design, and he was also the Lead Guest for a number of international journals.
Editor of the Journal of Communications, Special Issue on Network Coding
and Applications. He serves on TPCs of several international conferences
and workshops in the area of communications, networking, and statistical
signal processing.

ATHANASIOS MAGLARAS (M’06) was born in

Greece in 1948. He received the Electrical and
Mechanical Engineering degree, the M.Sc. degree,
and the Ph.D. degree from the National Technical
University of Athens, Greece, in 1971, 1985, and
2010, respectively. Since 1980, he has been a Pro-
fessor with the Technological Educational Institute
(T.E.I.) of Thessaly, Larissa, Greece. He was the
Head of the Electrical Engineering Department,
T.E.I. of Thessaly, for six years. He has been the
Head of the High Voltage and CAD/CAE Laboratory, T.E.I. of Thessaly,
since 1990. He has authored five books concerning high voltages, electric
fields, and CAD/CAE, and over 35 papers in international scientific maga-
zines and conferences, and meetings. The main research interests are high
voltages, electromagnetic field analysis, CAD/CAE, and electric vehicles.
He was a member of the Technical Chamber of Greece and the CIGRE. He
was a coordinator or a partner in many European research and educational
projects.

VOLUME 6, 2018 42565