A Survey of Networking Challenges and Routing Protocols in Smart Grids

Abstract

Smart grids (SG) represent the next step in modernizing the current electric grid. In this structure, a communications network is combined with the power grid in order to gather information that can be used to increase the efficiency of the grid, reduce power consumption, and improve the reliability of services, among other numerous advantages. SG communication networks are unique in their large scale and the limited capabilities of nodes which present several challenges in the design of efficient routing protocols. This paper provides a comprehensive survey of the main networking challenges present in the design of SG communication networks, and some of the important routing protocols proposed to address those challenges. Various technologies and architectures proposed for routing in SGs are discussed. A detailed comparison of the protocols considered in this paper is also given, and key areas that require further investigation are highlighted.

Full text

210 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 10, NO. 1, FEBRUARY 2014
A Survey of Networking Challenges and Routing
Protocols in Smart Grids
Ayman I. S a b b a h ,StudentMember,IEEE, Amr El-Mougy, and Mohamed Ibnkahla,M
ember, IEEE
Abstract—Smart grids (SG) represent the next step in modern-
izing the current electric grid. In this structure, a communications
network is combined with the power grid in order to gather infor-
mation that can be used to increase the efficiency of the grid, re-
duce power consumption, and improve the reliability of services,
among other numerous advantages. SG communication networks
are unique in their large scale and the limited capabilities of nodes
which present several challenges in the design of efficient routing
protocols. This paper provides a comprehensive survey of the main
networking challenges present in the design of SG communication
networks, and some of the important routing protocols proposed
to address those challenges. Various technologies and architectures
proposed for routing in SGs are discussed. A detailed comparison
of the protocols considered in this paper is also given, and key areas
that require further investigation are highlighted.
Index Terms—Networking, routing protocols, smart grid (SG).
I. INTRODUCTION
THE important concept behind smart grids (SGs) is inte-
grating two-way communications,sensing,andcontrol
technologies into the power system [1]–[4]. The major differ-
ence between telecommunication networks and the electric grid
is that communication networks route packets of information,
while the electric grid routes power flows. A telecommunica-
tion architecture for SGs has to encompass the telecommunica-
tion and power aspects of the system across all domains that may
be encountered. The National Institute of Standards and Tech-
nology (NIST) has created a high-level architectural model for
SGs [1]. In this model, the SG was divided into seven domains,
namely customers, markets, service providers, operations, bulk
generation, transmission, and distribution. This model serves as
a tool for identifying possible communication paths in SGs. It
is also useful for identifying intra-domain and inter-domain in-
teractions and potential applications enabled by these interac-
tions. The general SG architecture illustrates that different en-
tities within the SG are able to communicate with each other
via single or multiple hops. Routing protocols will determine
the paths needed for data flows. Therefore, the efficiency of the
telecommunications network, and thus the overall efficiency of
the SG, will depend on the utilized routing protocols.
The research in routing protocols for smart grids is scattered
over many books, journal papers, and conference papers. To the
Manuscript received March 16, 2012; revised May 31, 2012; accepted April
01, 2013. Date of publication April 18, 2013; date of current version December
12, 2014. Paper no. TII-12-0231.
The authors are with the Department of Electrical and Computer En-
gineering, Queen’s University, Kingston, ON Canada, K7M 3N6 (e-mail:
a.sabbah@queensu.ca; 6ema@queensu.ca; mohamed.ibnkahla@queensu.ca).
Digital Object Identifier 10.1109/TII.2013.2258930
best knowledge of the authors, there is no comprehensive survey
in the literature that compares between these protocols, identi-
fies their challenges, and addresses their capabilities to achieve
the end-to-end goals of the SG. Our main contribution in this
paper is to make an in-depth survey of existing protocols and to
present them in a unified framework that enables fair and crit-
ical comparison between them and address the need for new re-
search directions.
We have also identified areas of research that need further
investigation in this field and require specific attention from the
research community. In particular, this paper discusses different
SG applications and the inherent challenges and requirements to
support them.
Due to the importance of protocol evaluation, focus is also
given to the way that these protocols were evaluated. A compar-
ison of the evaluation techniques is provided and used to iden-
tify the applications that can be supported by each protocol.
The remaining sections of this paper are organized as follows.
In Section II, the challenges and requirements of SG routing pro-
tocols are discussed. In Section III, a taxonomy of routing pro-
tocols for SGs is presented. Section IV offers a comparison be-
tween the protocols, while Section V provides some concluding
remarks.
II. CHALLENGES,REQUIREMENTS,AND METRICS OFSG
ROUTING PROTOCOLS
One of the main steps in designing any routing protocol is to
study the traffic that will be supported. The SG is expected to
support a wide variety of applications. Section III describes the
main applications of the SG followed by the resulting challenges
and the routing metrics.
A. Applications and Challenges of SG
Several applications are expected to be supported by SG, each
having different Quality of Service (QoS) requirements. Some
applications and their requirements can be found in [5]–[7]. For
example, SG is expected to support some challenging appli-
cations such as video surveillance and voice data, as well as
several applications that are specific to the SG such as mon-
itoring the power quality (Advanced Metering Infrastructure
(AMI) [8]) and detecting power outages. It is also important to
note that almost all applications need security guarantees.
Thus, thefirst challenge for SG communications is to deter-
mine the optimum sensing, communications, and control hard-
ware. The coordination of centralized and distributed control
is another challenge [9]. Studies of the tradeoff between cen-
tralized and distributed control can be found in [10]–[12]. Fur-
thermore, routing protocols for SGs have to consider different
1551-3203 © 2013 IEEE

SABBAH et al.: SURVEY OF NETWORKING CHALLENGES AND ROUTING PROTOCOLS IN SMART GRIDS 211
types of traffic patterns, such as unicast, multicast, and broad-
cast communications, as well as different QoS requirements.
Special considerations must also be given to renewable energy
resources, which are expected to play key roles in SGs. How-
ever, it is important to note that most of these resources have
intermittent operation. Studies on the utilization of renewable
energy resources while considering intermittent operation can
be found in [13], [14].
In order to fulfill the requirements of SGs, certain routing
metrics are preferred. Some of these metrics include: packet la-
tency, Shortest Path (SP), packet delivery ratio (PDR), energy
metrics (e.g., node battery level), expected transmission count
(ETX), and some application-specific metrics that reflect the
QoS (e.g., jitter or data rate).
B. Communication Technologies in SG
There are several technologies under consideration for the
communication network of SGs. They can be classified into
wired or wireless technologies. Among the technologies pro-
posed for SGs is ZigBee [15], due to its capabilities for real-
time monitoring of multiple targets as well as self-organization,
self-configuration, and self-healing. This makes it for example
suitable to home area networks (HANs) applications. ZigBee
uses the unlicensed 2.4-GHz band for wireless communications,
which makes it vulnerable to interference from other technolo-
gies. In addition, ZigBee uses low data rates and the devices typ-
ically have limited memory. Alternatively, WiFi has sufficient
bandwidth for most applications but is a power-hungry standard.
On the other hand, there are several technologies under
consideration for the neighborhood area network (NAN) and
NAN-to-NAN (N2N) parts of the SG, where data is sent
from houses to utility centers. Among the wireless standards
considered for SGs are cellular networks (i.e., UMTS or LTE)
and wireless mesh networks (WMNs) such as WiMAX. Both
types of technologies have large bandwidth and are capable of
supporting various QoS requirements, making them suitable for
most SG applications. WMNs are capable of self-organization,
self-configuration, and self-healing and can transmit using
multihopping [16].
An alternative to wireless networks is to use the wired Pow-
erline Communications (PLC). There are several advantages for
using the power grid as a communication medium. First, power
cables readily exist, even in rural areas, and thus the cost of im-
plementing the communication network is low. Second, power
cables are owned by the utility company, which provide a de-
gree of independence from other networks. On the other hand,
there are some disadvantages for using PLC. Devices that are
turned off cannot be reached, which is an aspect that has to
be considered by routing protocols. Moreover, communication
signals that are transmitted through PLC channels experience
attenuation.
There are mainly two types of PLC communications [17]:
narrowband PLC (NBPLC), and broadband PLC (BPLC). Only
data rates in the order of a few tens of kbps can be achieved with
NBPLC [18]–[20], while BPLC can achieve data rates up to 100
Mbps. However, BPLC suffers from frequency-selective fading
over large distances. PLC can be used in HAN parts of the SG
whereitcanprovideareliablealternative in case of failure in
Fig. 1. Taxonomy of routing protocols for SG.
wireless links. Studies for the viability of using NBPLC in SG
can be found in [17] and [21].
C. Security in SG
Security is a critical component of SG and has to be de-
signed at the architectural level. Security includes the protection
from unauthorized accesses and malicious attacks. It also covers
the protection of compromised control units from harming the
system. The authors in [22] proposed a key management tool for
secure AMI. However, this paper dealt with the security issue
alone. Joint design of security and routing together is needed
for better efficiency. Routing data through the public SG com-
munications network is considered one of the weakest points in
the system and significant work is required in order to overcome
this vulnerability.
Security-aware solutions have to ensure that sufficient infor-
mation about a security event is available when and where it is
needed to support tactical decisions, such as preventing or min-
imizing disruption to SG services. This includes the collection
and delivery of the real-time data needed for situational aware-
ness [23]. Research should be focused on how to mitigate the
cyber security risks, which standards to use, and how to imple-
ment security mechanisms in future deployed grids [24]. A re-
view of some tools that can be used to provide security in SG
routing can be found in [25].
III. TAXONOMY OF ROUTING PROTOCOLS FOR SG
The protocols considered in this section are classified ac-
cording to the part of the SG they are most suitable for, partic-
ularly HAN, NAN, or N2N. Within each classification, several
routing methodologies are specified (e.g., topological specifica-
tions, metrics, or applications supported). Fig. 1 illustrates the
taxonomy used in this section.
A. Routing Protocols for HAN Parts of the SG
Routing protocols for HANs need to support low energy
consumption, ensure privacy and security of information, and
support self-organization and self-configuration in order to be
used easily by customers. To achieve these goals, the Internet
Engineering Task Force (IETF) proposed the Routing Protocol
for Low Power Lossy Networks (LLN), abbreviated as (RPL)
[26]. One of the advantages of RPL is that it is based on IPv6,

212 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 10, NO. 1, FEBRUARY 2014
Fig. 2. Operations performed by a meter node that has already joined the network and received a DIO message.
which supports multipoint-to-point (MP2P), point-to-multi-
point (P2MP), and point-to-point (P2P) communication, and
has enhanced capabilities for auto-configuration and security
[27]. RPL does not define any specific techniques for security,
other than the ones already available in IPv6 (these include
authentication and encryption techniques, as well as methods
to control the information sent over the network). An objective
function (OF) was introduced in RPL. Based on how the OF
is defined, this protocol should be able to support a variety of
applications and different QoS levels. For example, if the OF
combines the PDR metric with a metric that takes delay into
consideration, RPL will be a very good candidate to be used
in applications such as blackout area detection and monitoring
and substation and transformer monitoring. RPL constructs a
directed acyclic graph (DAG) with one or more sink nodes
as its roots. The DAG was designed to minimize the cost of
reaching its roots, as specified by the OF, from any node in the
network. RPL also employs the trickle timer technique [28] that
adapts traffic in order to maintain stability and load balancing.
In order to discover the routing paths, the root of the DAG
starts sending out DAG information option (DIO) messages pe-
riodically. When a node broadcasts its DIO message, it includes
information about its rank (the distance of the node from the
backbone network), the OF, DAG-ID it has joined in, among
others. Any node that is connected to the backbone network can
act as a root or LLN border router (LBR) and has a rank equal to
1. Once a node receives a DIO, it calculates its rank based on the
rank in the received DIO, and the cost of reaching the root node
from itself. It also selects its parent based on the local quality of
the links, the advertised OF, path cost, and rank, among others.
Any node that has lower rank than the node itself is considered
as a candidate parent. The flowchart shown in Fig. 2 describes
the RPL algorithm for AMI [29].
The performance of RPL was studied in [29]–[32]. In [30],
some modifications were proposed for rank computation and
path recording in order to enable real-time meter reading and
service management. The modified system was shown to out-
perform the Ad-hoc On-demand Distance Vector (AODV) pro-
tocol in terms of latency. AODV is considered a reactive routing
protocol. It uses flooded RREQ messages to discover paths to
destinations on demand. A performance comparison between
RPL and a geographical routing protocol for AMI was pre-
sented in [31]. The routing mechanism of such protocols incor-
porates geographical coordinates of the nodes in making routing
decisions. The geographical routing protocol adopted a com-
bination of weighted link metrics and distance vectors, while
RPL utilized the ETX metric. Path loss, temporal fading, and
log-normal shadowing can be considered when calculating the
links’ status.
ApracticalsolutionbasedonRPLwasproposedin[32]to
support AMI mesh networks where it was suggested to divide
the network into multiple subnetworks. This protocol was eval-
uated using testbed measurements. Delay and PDR were used
for evaluation. Findings demonstrate that the suggested solution

SABBAH et al.: SURVEY OF NETWORKING CHALLENGES AND ROUTING PROTOCOLS IN SMART GRIDS 213
might improve scalability and performance of the network. It
was found that nodes that are one hop away attain the highest
PDR. Studying the proposed algorithm for different and large-
scale topologies is required as well as comparing the perfor-
mance with other algorithms. A study of routing metrics that
can be used in RPL can be found in [33]. To the best knowledge
of the authors, there is no work on how to make the OF adaptive
so that it supports the requirements of different applications.
On the other hand, the idea of using PLC to connect sensor
networks in HAN based on IPv6 was investigated in [34].
Communications were based on a modified version of the
IEEE Standard 802.15.4 due to its ability to support IPv6
in Low-power wireless personal area networks (6LoWPAN)
[35], [36]. Watt pulse communications (WPC) was used [37],
which exploits the transient behavior of electrical networks to
generate narrow pulses that carry information. A comparison
between 6LoWPAN and WPC was presented in [34], which
showed that WPC supports lower data rates.
In order to enable interoperability between PLC and
6LoWPAN networks, a multiphysical layer protocol stack,
called M-Stack, was proposed in [38]. Five network imple-
mentations were provided based on the RPL protocol. One of
the networks was based on PLC and the others were based on
the wireless IEEE Standard 802.15.4. Each network had its
own hardware implementation. The bridge was made at the IP
level by a router where an RPL implementation was running.
ETXwasusedastheroutingmetric. Also, only ping packets
were used to test the architecture. A testbed was used, which
showed that nodes from all of the wireless networks were
pinged successfully from the PLC nodes and vice versa. This
testbed showed that this M-stack protocol might be a possible
solution for heterogeneity. Moreover, a comparison of latency
vs. payload was performed for PLC network using simple
no-hop and one-hop topologies. The improvement was about
15% in no-hop topology and 17% in one-hop topology.
The authors in [39] study the interoperability between PLC
and wireless networks. A complete networking architecture
based on DAGs was introduced, labeled Hybrid-LLN. In this
work, three types of nodes were considered: PLC nodes, wire-
less nodes, and hybrid nodes that support PLC and wireless
communications. The metric selected in this case was the
estimated energy of each node. Thus, the path that wasted the
least energy was chosen. Based on [40], every battery-powered
node should be one hop away from a line-powered node in
order to achieve maximum network lifetime. This topology
was applied in [39] and the results showed that mixing PLC
with wireless communications in the same topology increased
the lifetime significantly; since wireless/PLC nodes reduce the
load on battery-operated wireless nodes, this leads to balancing
node energy consumption across the network [41].
B. Routing Protocols for NAN Parts of the SG
NAN parts of the SG carry information from meters to the
utility centers. Thus, these networks must have sufficient ca-
pabilities to support the different QoS requirements of users.
One of the main protocols considered for NAN is the RPL pro-
tocol [26], due to its capabilities in supporting multiple com-
munication types, multiple routing metrics (in the OF), as well
Fig. 3. Average data packet delay per node in DADR protocol.
as security. Another routing protocol called HYDRO was in-
troduced in [42] for P2P wireless communications. HYDRO is
a hybrid-routing protocol for LLNs that provides both central-
ized control and local agility. It used a distributed algorithm to
form a DAG for routing data from in-network nodes to border
routers, similar to the method used by RPL. Nodes were allowed
to maintain multiple options for routing. Topology reports were
piggybacked on periodic traffic, thus allowing border routers
to build and maintain a global view of the network. The DAG
provided P2P routing by allowing low-power nodes to forward
packets to a border router, which in turn routes them to the ap-
propriate destination. HYDRO has a centralized optimization
feature that enables border routers to insert routing table en-
tries at appropriate nodes, which increases the reliability of the
chosen paths. Control traffic statistics were used as the only
metric to test HYDRO. With the expected high reliability level
of this protocol, due to using multiple and alternative routes,
it might be suitable for applications such as monitoring power
quality. However, other metrics need to be taken in considera-
tion to test the ability of HYDRO to support various applica-
tions. Moreover, this protocol gives no consideration for secu-
rity aspects.
In [43], the distributed autonomous depth-first routing
(DADR) protocol was proposed, which was composed of two
mechanisms. The first mechanism was a light weight control
plane used to keep a soft routing table with redundant paths for
each node. The second mechanism was a forwarding plane that
had the capability of routing data packets by doing a depth-first
search guided by the routing table. The advantages of the
proposed algorithm are significant if the topology is dynamic.
In this case, instead of flooding the network to search for new
routes, the data forwarding plane used redundant routes from
the routing table to forward the packets. This makes DADR
very useful for tracking EVs and moving control units. DADR
routing technology was implemented on IEEE 802.11, IEEE
802.15.4 and IEEE 802.3. During routing initialization and
paths discovery, the only control message used was a periodic
HELLO message sent every minute. This causes time delays
that may have significant effects on the system performance.
Fig. 3 shows that the worst delay experienced at a node is
1.2 s, while the majority of the packets were delivered in less
than 0.5 s, which may be unsuitable for some applications that
require strict low-delay requirements.

214 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 10, NO. 1, FEBRUARY 2014
Fig. 4. Standard HWMP-based routing procedure.
In [44], a WMN based on the IEEE Standard 802.11s was
proposed as a high-speed backbone network for SG. This is be-
cause it supports the hybrid wireless mesh protocol (HWMP),
which is considered suitable for static mesh networks. NAN and
HAN zones of SG are examples of such networks. IEEE 802.11s
provides some schemes to prioritize data traffic to support high
QoS for time-critical data. In addition, unicast, multicast, and
broadcast communications are supported. Fig. 4 shows the stan-
dard HWMP-based routing procedure. However, HWMP might
not be fully suitable for SG. This is due to route instability that
might occur. Also, the used routing metric is the link cost which
might not fit SG very well. Thus, a modified version of HWMP
was proposed in [44]. In this version, two types of routing algo-
rithms were utilized: tree-based and on-demand routing.
The Root Announcement mechanism (RANN) was used,
whereby the gateway broadcasts messages so that each node
could calculate airtime cost metrics representing latency and
error rate of a specific path. This metric was defined as
where is the channel access overhead, is the size of the
transmission frame, is the data rate, and is the error rate.
The information about the links was calculated at each node
and accumulated inside the received RANN message. Due to the
fact that SG will have different types and sizes of packets, new
ways to calculate the error rate were needed. It was suggested
to use the MAC retransmission rate to calculate the failure rate
while accounting for the variable size of packets. The calcula-
tion was performed as follows:
where is the total number of MAC-level retransmissions
made by node ,is the total number of packets transmitted
Fig. 5. Performance evaluation of original and modified HWMP protocols
[44]. (a) PDR. (b) End-to-end delay.
by node ,is the maximum allowed retransmissions,
is the size of the packet ,is the largest size of a packet
in the network. A comparison between the performance of the
proposed version and original HWMP is shown in Fig. 5.
In the modified HWMP, multiple routes were allowed to be
maintained at each node. This reduced route fluctuation and
improved the performance of the network. However, a large
amount of information had to be maintained at nodes, which
affects complexity, node lifetime, and other factors. The eval-
uation for the proposed protocol was performed for 36 nodes.
The topology used was simple and is not representative of a real
SG. Moreover, the results showed that congestion occurs when
the network contains 25 nodes or more. Thus, there is a need to
optimize the location of gateways and deploy routers to prevent
congestion. This protocol might be suitable for applications that
tolerate some packet loss such as monitoring peak rate and time.
Another algorithm that is based on HWMP was proposed re-
cently in [45]. Some modifications were proposed in order to
enhance the routing reliability when using WLAN as a back-
bone for SG in NANs. This protocol is designed to improve the
stability of HWMP routing protocol that might occur when the
routing path of a node changes constantly during data transmis-
sion. In order to achieve this, a new method for calculating air-
time cost metric for the HWMP of 802.11s standard was pro-
posed. In order to prevent packet loss during link breakage, a
resource reservation algorithm was proposed. Moreover, a la-
tency-tolerant traffic management scheme was investigated in
order to prioritize SG data when using HWMP. Simulation re-
sults showed that these modules may provide highly reliable
data transmission as well as prioritization of data in order to
support various SG services.

SABBAH et al.: SURVEY OF NETWORKING CHALLENGES AND ROUTING PROTOCOLS IN SMART GRIDS 215
Due to the importance of the security issue, some routing
protocols were designed with the primary objective of guaran-
teeing security in wireless SG communication networks. For ex-
ample, a secure routing protocol called Secure Routing Archi-
tecture (SRA) was proposed in [46]. This protocol assumed a
hierarchical architecture for SG, and employed a hybrid control
scheme between multiple levels of the hierarchy. Centralized
control was employed at the higher levels since the number of
nodes decrease with higher levels of the hierarchy. Thus, a cen-
tralized control scheme can provide more optimized decisions
at critical data centers. On the other hand, a distributed control
scheme was employed at lower levels of the hierarchy due to
the large number of nodes. It was suggested in [46] that there is
a tradeoff between security and latency, since more security im-
plies more redundancy and processing, and therefore longer de-
lays. Thus, a game-theoretic model was proposed to find a com-
promise between security and latency [46]. However, no evalu-
ation was done for this protocol neither by simulation nor using
a testbed. Also, there was no consideration for multiple QoS re-
quirements for different applications. Generally speaking, this
protocol might be suitable for power outage detection because
it requires a high level of security, reliability, and a controllable
level of latency that can serve real-time applications.
In [47], a WMN system architecture called Geo-Mesh was
proposed for energy management applications in NAN. Here
geographical routing [77] was adopted with a combination of
weighted link metrics and geographical proximity. P2P com-
munications was supported, where nodes chose the neighbor
which was geographically closest to the final destination using
knowledge of the node coordinates. If there was a router in
the neighborhood, the node chose it as its parent regardless of
other considerations since it had higher communication capa-
bilities. However, the routing protocol suggested that metering
data be pushed to the sink once every 4 hours which is not
practical for most SG applications. For security, the system em-
ployed the Public Key Infrastructure (PKI) architecture to pro-
tect against cyber attacks. There was no study as to how geo-
graphical routing might guarantee the QoS requirements of dif-
ferent SG applications.
A Secure Energy Routing Mechanism (SERM) protocol was
proposed in [48]. Here, message redundancy is used to detect
internal attacks such as spoofed route signaling and fabricated
routing messages. SERM contains secure key management and
route discovery based on energy metrics. Computer simulations
were carried out to evaluate the performance of SERM. 100
nodes were deployed in a 3000 m 3000 m square field, and
five nodes were selected to send packets every 20 min. The pres-
ence of malicious routers that send unencrypted packets was
assumed. The results showed that PDR decreases and delay in-
creases as the number of malicious routers increase. SERM was
able to separate the malicious data and prevent them from prop-
agating across the network.
Several routing protocols based on PLC communications
were also proposed for NAN. For example, the Hybrid-LLN
protocol [39] that was previously reviewed in Section III-A
is also suitable for NAN. In addition, the project known as
Remote Energy Management over Power Lines and the Internet
(REMPLI) [49] targeted the implementation of a SG using PLC
in some areas of Europe, where NBPLC solutions are used
[50]. It was mainly designed to support energy management
and AMI. Meters were equipped with a PLC modem/repeater
and data was aggregated at a central sink located at the medium
voltage/low voltage (MV/LV) transformer. The nodes formed
a tree mesh network. The master-slave communication model
was adopted by REMPLI and two routing protocols were
proposed. The first one was called DLC1000 [51] and used dy-
namic source routing. The master nodes used periodic polling
to maintain routing tables that specified the best paths to reach
the slave nodes. The second protocol used a Single Frequency
Network (SFN) [52], [53]. SFN employs tree-based hierarchy
and flooding-based routing to transmit packets between master
and slave nodes.
In order to compare the performance of DLC1000 and SFN,
an analytical model was developed in [54]. Several network
topologies were considered, particularly random, ring, and tree
topologies. The maximum number of nodes used was 200. The
authors in [54] studied the bandwidth consumed and the time
taken for a master node to poll all of the slave nodes for each
protocol. The results showed that DLC1000 incurred more de-
lays and consumed more bandwidth for routing overhead than
SFN. In addition, sometimes a few slave nodes could not be
reached with DLC1000. Thus, it was concluded that SFN per-
forms better than DLC1000. Note that, in DLC1000, the master
maintains the routing table by periodically getting information
about the paths from slaves. This information is then used to
determine the best path for transferring a packet to a slave. On
the other hand, in SFN, the master stores a table that specifies
the number of repeaters necessary for reaching every slave (re-
peater level). When a slave receives a packet correctly, it checks
its header and if the remaining repeater level is zero it does not
transmit this packet, otherwise, it continues to retransmit the
packet and decrease the repeater level by 1. Because power grid
infrastructure is rather static, which means that the locations of
most nodes are fixed, using geographical routing protocols for
PLC might be suitable. The geographical protocol proposed in
[55] considered the use of PLC in low- and medium-voltage dis-
tribution grids to connect network nodes (e.g., meters, actuators,
sensors) through multihop transmissions. The nature of this pro-
tocol makes it suitable for AMI and demand optimization appli-
cations, specifically for transmitting the control messages from
the head node to different appliances at homes. It is also helpful
in aggregating the meters’ readings to the utility centre.
Cooperation in communication is also another idea for im-
proving the system efficiency. The authors of [56] introduced
and compared different approaches of cooperative multihop
transmission in NBPLC networks. They also presented different
schemes for coding and relay selection in multihop transmis-
sion. The authors concluded that no diversity gain can be
achieved from cooperative multihop communications but it
provides better rate gains as compared with conventional
multihop communications.
C. Routing Protocols for N2N Parts of the SG
At the highest level of the hierarchy of SG communication
networks lies the high bandwidth backbone that can transport
large volumes of data with strict QoS requirements. WMN and

216 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 10, NO. 1, FEBRUARY 2014
Fig. 6. Routing procedure for the Multi-Mesh protocol.
BPLC are the main candidates for communication in this part of
the SG. In [57], Multi-gate Mesh (Multi-Mesh) network archi-
tecture was proposed for NAN and N2N zones of the SG. Multi-
gate routing, multichannel routing, and real-time traffic sched-
uling techniques were proposed. IEEE 802.11s was used as the
core routing protocol, and an extension was added to support the
multi-gate network architecture. Two types of nodes were pro-
posed, the first new node is called data aggregator point (DAP),
which acts as the neighborhood gateway node; while the second
type was called master-gateway station (MGS), which acts as
AMI head-end and is connected to the backbone. DAP periodi-
cally broadcasts root announcement messages, which allow me-
ters to discover the address of the transmitting node. The routing
procedure for Multi-Mesh is shown in Fig. 6. This protocol is
suitable for home energy management (HEM) [58] and AMI
applications.
In [59], a fault-tolerant routing method called FT-Overlay
based on multiple paths was presented. The idea was to pro-
vide a set of links, called crosslinks, placed at different parts
of the network to be used in fault scenarios. The number of
crosslinks depends on the height ( ), which represent the
levels of hierarchy in the network. The number of crosslinks
is found using: . Two approaches were proposed.
In the first one, devices were divided into partitions based
on their IP addresses. These partitions were then grouped in
clusters of fixed size, which were then interconnected with
crosslinks to provide redundant communication pathways. A
weakness of this approach is that it depends on an overlay
structure even if there are no faults in the network, which
will needlessly decrease throughput. In the second approach,
static routing was used and crosslinks were utilized in case of
fault occurrence only. The reliability of this protocol makes
it very suitable for the applications such as power quality
monitoring and blackout detection.
D. Heterogeneous Solutions for SG
1) Applicability of WSN Technology in SG: The leading
technology in WSN is the ZigBee protocol stack, which is based
on the MAC and physical layers of the IEEE Standard 802.15.4.
Advanced algorithms and protocols have been proposed re-
cently (such as cognitive routing [77] and weighted cognitive
maps [78]). This allows WSN achieve end-to-end goals while
fulfilling application requirements such as power, delay, and
scalability. ZigBee nodes cannot directly communicate with
servers outside the WSN if they are using other standards. Thus,
if there is a need to remotely control nodes (which is the case
in SG), an additional mechanism is required. A gateway can be
deployed to mediate between the nodes and the servers on the
Internet [15]. In [60], a system architecture that uses Wi-Fi was
proposed as a gateway for home automation using ZigBee. In
[61], a study regarding challenges and opportunities of using
WSN in SG was presented. As mentioned before, ZigBee
also provides some security measures such as encryption and
authentication. There is no perfect solution for interconnecting
ZigBee and IP.
While routing is performed in the network layer in IP net-
works, an adaptation layer is needed for 6LoWPAN in order to
enable the transmission of IPv6 packets over IEEE 802.15.4
technology. In the migration period, it is important to ensure
interoperability between ZigBee and IP technologies (e.g.,
6LoWPAN). An attempt for studying interoperability between
ZigBee and 6LoWPAN for SG applications was presented in
[62] using Session Initiation Protocol (SIP).
2) Interoperability in SG: In previous sections, some en-
abling technologies for SG communications networks, such
as NBPLC, BPLC, WMN, and WSN, have been discussed.
Each option has its advantages and disadvantages. NBPLC is
less susceptible to fading effects than BPLC but it achieves
lower data rates. BPLC achieves higher data rates but suffers
from frequency-selective fading over larger distances. BPLC
also requires the implementation of couplers, making it more
expensive. Both NBPLC and BPLC suffer from noise pulses
and the open circuit problem (devices becoming unreachable)
due to switching devices ON/OFF [63]. A comprehensive study
of PLC channel characteristics can be found in [64] and [65].
Wireless networks are an alternative that can achieve high
data rates and are independent from the power grid topology.
WMN is one of the options proposed for implementing wire-
less SG networks due to its ability to cover large areas and
achieve high data rates. This makes it a good candidate as a high-
speed backbone for the SG [66]. WSN/ZigBee technologies pro-
vide an efficient solution for low-powered devices, making them
suitable for smart home applications [67].
Single communications technologies may not be suitable for
all SG areas and applications. Integrating some of these tech-
nologies may be advantageous. However, there is still a debate
on which technology to use [68]–[70]. Several studies [38], [39],
[41], [71] have shown that integrating PLC and wireless tech-
nologies improves the efficiency of the SG in terms of the com-
munication link quality. In a hybrid network, interoperability is
a major concern and gateways have to be used when needed.

SABBAH et al.: SURVEY OF NETWORKING CHALLENGES AND ROUTING PROTOCOLS IN SMART GRIDS 217
TAB L E I
COMPARISON OF QOS ROUTING PROTOCOLS
IV. COMPARISON BETWEEN SG ROUTING PROTOCOLS
Here, a comparison between some of the protocols discussed
in Section III is presented. The comparisons are based on the
routing methodology used, performance evaluation, and the ap-
plications supported by the protocols. Table I outlines the main
features of each protocols.
A. Routing Methodology
The protocols surveyed in Section III use different routing
methodologies. As Table I shows, none of the protocols con-
sidered is purely centralized. All protocols are either fully dis-
tributed or have a hybrid architecture of centralized and dis-
tributed control. This is due to the large size of the SG, which
makes it difficult for a single controller to manage the entire net-
work. A clustered architecture, where every group of nodes is
managed by a cluster head, may be suitable for SG networks as
it can exploit the heterogeneity of nodes.
In addition, not all protocols support different traffic patterns
(e.g., P2P, P2MP, and MP2P). This is critical in SG since the
three modes of communication will probably coexist in the
network. More research is needed to study network perfor-
mance under multimode communications. For example, P2MP
and MP2P utilize multiple paths simultaneously, which may
increase interference and possibly cause congestion. Also,
few protocols consider QoS routing metrics directly. Most
protocols shown in Table I consider ETX or SP in discovering
paths. These metrics may not be efficient in providing guar-
antees for requirements such as delay or PDR. Also, routing
protocols should be able to support real-time and nonreal-time
communications. The Multi-mesh protocol [59] includes traffic
scheduling for QoS support.
Protocols that use IPv6 [25], [38] can exploit its prioritiza-
tion functions. Most protocols use either DAG or tree-based
routing. These methods require less overhead than on-demand
routing and are suitable for static networks such as SGs.
On-demand routing was coupled with tree-based routing in
[44] and distance-vector routing in [43] to support dynamic
topologies. Geographical routing was used in [48], where paths
can be easily discovered using node coordinates. However,
geographical routing requires either the addition of GPS re-
ceivers or hard-coding coordinates at every node. GPS devices
are expensive, and hard-coding coordinates will mean that the
locations of devices cannot be changed. However, there are
some challenges to geographical routing in SGs. For example,
consumers relocating to other houses may have to re-encode
all their appliances. Furthermore, for apartment buildings
and dense urban areas, determining the geographical location
becomes a challenging task [77].
Route maintenance is critical in SGs in order to enable
self-healing of faults. All protocols shown in Table I employ
route maintenance mechanisms. The protocols in [32], [43],
[57], and [59] establish multiple routes that can be used alter-
natively when a route fails. This method is suitable in DAG

218 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 10, NO. 1, FEBRUARY 2014
TAB L E II
PERFORMANCE EVA L U AT I O N O F ROUTING PROTOCOLS
and tree-based routing since multiple entries may readily exist
in the routing table. In addition, only protocols in [25], [38], [39]
employ proactive route maintenance, where links or nodes are
replaced before they fail. This can have a significant impact in
guaranteeing QoS.
B. Protocol Evaluation and Applications Supported
It was mentioned in Section III that several protocols were
not thoroughly evaluated. Therefore, in order to identify ways
in which protocol evaluation can be improved, a comparison

SABBAH et al.: SURVEY OF NETWORKING CHALLENGES AND ROUTING PROTOCOLS IN SMART GRIDS 219
between the evaluation techniques of some of the surveyed pro-
tocols is provided in Table II.
Table II shows that the maximum number of nodes that was
used in evaluating any of the listed protocols was 2607 nodes
[43]. A typical SG that covers an entire city may contain thou-
sands of nodes. Thus, the simulations may not indicate how pro-
tocols will perform in a real-world implementation. We can no-
tice from Table II that delay was the main metric under consid-
eration when evaluating most protocols. Although delay is of
primary importance in SGs, other metrics need to be considered
as well to give a clear picture about QoS performance.
In addition, none of the protocols tested the coexistence of
different traffic patterns, which is an important issue, as men-
tioned before. It would also be interesting to see evaluations
using real information about the energy usage of customers,
since this can have an impact on the nature of the trafficand
its intensity. This was explored in [43], and should be extended
to other research efforts. It is difficult to determine the specific
applications that can be supported by each protocol. This is be-
cause all protocols have been evaluated for small sets of metrics.
For example, Table II shows that all of the listed protocols have
considered AMI readings. Although AMI reading is an impor-
tant application in SG, evaluating the protocols for other appli-
cations needs to be explored as well. Table II shows that RPL
outperforms AODV and geo-routing in terms of delay. Thus,
RPL may be suitable for real-time applications such as outage
detection and power quality monitoring. However, these results
give no indication as to the level of reliability that RPL may
guarantee. Thus, we do not know how RPL may perform for
applications that require strict levels of PDR. Also, DADR was
the only protocol that supported mobility. This may be critical
for applications such as monitoring electric vehicles.
It can be concluded that new routing protocols must be de-
signed with various applications in mind, and performance eval-
uation has to be more comprehensive. Finally, it is important to
note that some projects for testing and deployment of SGs in
real-world conditions are on the way [72]–[76]. For example,
Mitsubishi Electric has recently started full-scale tests in Japan
on how to manage a large number of renewable energy resources
in a power grid [75]. General Motors (GM) and OnStar have
launched a project for managing the operation of electric vehi-
cles [76]. Data collected from cars will be used by the utility
company to optimize the time for charging the batteries in order
to minimize the electricity bills for drivers. The data will also be
used to reduce peak power demand by shifting battery charging
to off-peak hours. Extensive details of these projects are beyond
the scope of this paper.
V. C ONCLUSION
In this paper, a comprehensive survey of routing protocols
for SG communications was presented. The challenges and re-
quirements for SG communications were discussed, as well as
possible metrics that may be considered. Protocols that use wire-
less and PLC communications were illustrated, as well as ideas
from WMNs and WSNs.
A detailed comparison between the protocols discussed in
this paper was provided. In addition, areas that require further
research were highlighted. For example, there is a clear need
for a more comprehensive design of SG protocols, due to the
large size of the network and the multiple challenges present.
The network layer and routing protocols have to be designed
with a network management perspective. Thus, the protocols
have to be aware of the status of the nodes (e.g., their available
capabilities and resources) and the requirements of the targeted
applications. Also, the option to integrate PLC communications
within the network may be highly valuable. Interoperability is-
sues and different capabilities of the mediums have to be con-
sidered in the routing protocol. Multi-hopping in PLC systems
also needs to be studied further. In order to guarantee interop-
erability, a new protocol stack may need to be developed. This
is also important for the utilization of different technologies in
the SG such as ZigBee and WMN.
REFERENCES
[1] “NIST Framework and Roadmap for Smart Grid Interoperability Stan-
dards, Release 1.0,” Nat. Inst. Standards Technol., Gaithersburg, MD,
USA, 2010 [Online]. Available:http://www.nist.gov/public_affairs/re-
leases/upload/smartgrid_intEroperability_final.pdf
[2] M. Kintner-Meyer, K. Pratt, and R. Schneider, Impact Assessment of
Plug-In Hybrid Vehicles on Electric Utilities and Regional U.S. Power
Grids Part 1: Technical Analysis. Richland, WA, USA: Pacific
Northwest Nat. Lab., U.S. DOE, 2006.
[3] “The Smart Grid: An Introduction,” US DOE, 2008 [Online]. Avail-
able: http://energy.gov/oe/downloads/smart-grid-introduction-0
[4] G. Arnold, “Challenges and opportunities in smart grid: A position ar-
ticle,” Proc. IEEE, vol. 99, no. 6, pp. 922–927, Jun. 2011.
[5] B. Flynn, Key Smart Grid Applications [Online]. Available:
http://www.smartgridnews.com/artman/uploads/1/Smart_Grid_Appli-
cations.pdf
[6] P. Palensky and D. Dietrich, “Demand side management: Demand re-
sponse, intelligent energy systems, and smart loads,” IEEE Trans. Ind.
Inf., vol. 7, no. 3, pp. 381–388, Aug. 2011.
[7] Dept. Energy, Commun. requirements of Smart Grid technologies,
Oct. 2010 [Online]. Available: http://energy.gov/sites/prod/files/
gcprod/documents/Smart_Grid_Communications_Requirements_Re-
port_10-05-2010.pdf
[8] F. Benzi, N. Anglani, E. Bassi, and L. Frosini, “Electricity smart meters
interfacing the households,” IEEE Trans. Ind. Electron.,vol. 58, no. 10,
pp. 4487–4494, Oct. 2011.
[9] A. Vaccaro, G. Velotto, and A. Zobaa, “A decentralized and cooperative
architecture for optimal voltage regulation insmart grids,” IEEE Trans.
Ind. Electron., vol. 58, no. 10, pp. 4593–4602, Oct. 2011.
[10] G. Zhabelova and V. Vyatkin, “Multiagent smart grid automation ar-
chitecture based on IEC 61850/61499 intelligent logical nodes,” IEEE
Trans. Ind. Electron., vol. 59, no. 5, pp. 2351–2362, May 2012.
[11] P. Levis, A. Tavakoli, and S. Dawson-Hag, “Protocols for low power
and lossy networks,” in Internet Engineering Task Force, Internet
draft-IETF-ROLL-protocols-survey-07, 2009.
[12] H. Kanchev, F. Di Lu, V. Lazarov, and B. Francois, “Energy manage-
ment and operational planning of a microgrid with a PV-based active
generator for smart grid applications,” IEEE Trans. Ind. Electron., vol.
58, no. 10, pp. 4583–4592, Oct. 2011.
[13] C. Cecati, C. Citro, A. Piccolo, and P. Siano, “Smart operation of wind
turbines and diesel generators according to economic criteria,” IEEE
Trans. Ind. Electron., vol. 58, no. 10, pp. 4514–4525, Oct. 2011.
[14] C. Molitor, A. Benigni, A. Helmedag, K. Chen, D. Cali, P. Jahangiri,
D. Muller, and A. Monti, “Multi-Physics testbed for renewable energy
systems in smart homes,” IEEE Trans. Ind. Electron., vol. 60, no. 3,
pp. 1235–1248, Mar. 2012.
[15] “Understanding ZigBee Gateway,” ZigBee Alliance, ZigBee Docu-
ment 095465r13, Sep. 2010.
[16] I. Akyildiz and X. Wang, “A survey on wireless mesh networks,” IEEE
Commun. MAg., vol. 43, no. 9, pp. S23–S30, Sep. 2005.
[17] F. Aalamifar, “Viability of powerline communications for smart grid
realization,” M.A.Sc. thesis, Dept. Electr. Comput. Eng., Queen’s
Univ., Kingston, ON, Canada, 2012.
[18] M. Bauer and M. Kruse, “Countering impulsive noise without channel
coding,” in Reports on Industrial Information Technology.Karl-
sruhe, Germany: KIT Scientific, 2010, vol. 12, pp. 95–112.

220 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 10, NO. 1, FEBRUARY 2014
[19] Signaling on Low Voltage Electrical Installations in the Frequency
Range 3 kHz to 148.5 kHz, CENELEC EN 50065, CENELEC, 2002.
[20] Standard for Low Frequency (Less Than 500 kHz) Narrow Band Power
Line Communications for Smart Grid Applications, P1901.2, 2010,
IEEE standard.
[21] M. Sigle, M. Bauer, L. Wenqing, and K. Dostert, “Transmission
channel properties of the low voltage grid for narrowband power line
communication,” in Proc. IEEE Int. Symp. Power Line Commun. Its
Applications, Apr. 3–6, 2011, pp. 289–294.
[22] N.Liu,J.Chen,L.Zhu,J.Zhang,andY.He,“Akeymanagement
scheme for secure communications of advanced metering infrastruc-
ture in smart grid,” IEEE Trans. Ind Electron., vol. 60, no. 10, pp.
4746–4756, Oct. 2012.
[23] H. Kenchington, “Securing tomorrow’s grid (Part I),” Public Utilities
Fortnightly, vol. 149, no. 7, pp. 28–42, 2011.
[24] H. Kenchington, C. Hawk, D. R. Highfill, J. Eisenhauer, and L. K ishter,
“Securing tomorrow’s grid (Part II),” Public Utilities Fortnightly, vol.
149, no. 8, pp. 42–48, 2011.
[25] A. Wrigh, P. Kal, and R. Siber, “Interoperability and security for
converged smart grid networks,” in Proc. Grid-Interop Forum,
2010 [Online]. Available: http://www.gridwiseac.org/pdfs/forum_pa-
pers10/wrigth_gi10.pdf
[26] T.Winter,P.Thubert,A.Brandt,T.Clausen,J.Hui,R.Kelsey,P.
Levis, K. Pister, R. Struik, and J. Vasseur, “RPL: IPv6 routing pro-
tocol for low power and lossy networks,” in Internet Engineering Task
Force, Internet draft-IETF-roll-RPL-19, Mar. 2011.
[27] J. W. Hui and D. E. Culler, “IP is dead, long live IP for wireless sensor
networks,” in Proc. 6th ACM Conf. Embedded Network Sensor Syst.,
2008, pp. 15–28.
[28] P. Levis, E. Brewer, D. Culler, D. Gay, S. Madden, N. Patel, J. Po-
lastre,S.Shenker,R.Szewczyk,andA.Woo,“Theemergenceofa
networking primitive in wireless sensor networks,” Commun. ACM,
vol. 51, no. 7, pp. 1–9, Jul. 2008.
[29] J. Tripathi, J. de Oliveira, and J. Vasseur, “Applicability study of RPL
with local repair in smart grid substation networks,” in Proc. IEEE Int.
Conf. Smart Grid Commun., Oct. 2010, pp. 262–267.
[30] D. Wang, Z. Tao, J. Zhang, and A. Abouzeid, “RPL based routing for
advanced metering infrastructure in smart grid,” in IEEE International
Conference on Communications (ICC), May 2010, pp. 1–6.
[31] G. Iyer, P. Agrawal, E. Monnerie, and R. Cardozo, “Performance
analysis of wireless mesh routing protocols for smart utility net-
works,” in Proc. IEEE Int. Conf. Smart Grid Commun.,Oct.2011,
pp. 114–119.
[32] P. Kulkarni, S. Gormus, F. Zhong, and F. Ramos, “AMI mesh net-
works—A practical solution and its performance evaluation,” IEEE
Trans. Smart Grid, vol. 3, no. 3, pp. 1469–1481, Sep. 2012.
[33] J. Vasseur, M. Kim, K. Pister, and H. Chong, “Routing Metrics Used
forPathCalculationinLowPowerandLossyNetworks,”Draft-IETF-
roll-routing-metrics-04, Dec. 2009.
[34] C. Chauvenet, B. Tourancheau, and D. Genon-Catalot, “802.15.4, a
MAC layer solution for PLC,” in Proc. IEEE Int. Conf. Comput. Syst.
Applications, May 2010, pp. 1–8.
[35] G. Montenegro, N. Kushalnagar, J. Hui, and D. Culler, “IPv6 over low
power wireless personal area networks (6LoWPAN),” IETF RFC 4944,
2007.
[36] G. Montenegro, N. Kushalnagar, J. Hui, and C. Culler, “Transmission
of IPv6 packets over IEEE 802.15.4 networks,” IETF RFC 4944, 2007.
[37] “WPC Functioning,” Watteco, Oct. 2012 [Online]. Available:
http://www.watteco.com/index.php/solutions/6lowpantransceivers/
wpc-functionning.html
[38] C. Chauvenet, B. Tourancheau, D. P. Ganon-Catalot, P.-E. Goudet, and
M. Pouillot, “A communication stack over PLC for multi-layer phys-
ical IPv6 networking,” in Proc. IEEE Int. Conf. Smart Grid Commun.,
Oct. 2010, pp. 250–255.
[39] L. B. Saad, C. Chauvenet, and B. Tourancheau, “Heterogeneous IPv6
infrastructure for smart energy efficient building,” in Proc. Dubrovnik
Conf. Sustainable Development of Energy, Water and Environment
System, 2011, pp. 1–13.
[40] M. Yarvis, A. Kushalnagar, H. Singh, Y. Liu, and S. Singh, “Exploiting
heterogeneity in sensor networks,” Proc. IEEE INFOCOM, vol. 8, pp.
878–890, Mar. 2005.
[41] K. Yamada, Y. Hirata, Y. Naoe, T. Furumura, Y. Inoue, T. Shimizu,
K. Yoshida, M. Kojima, and T. Mizuno, “Dual communication
system using wired and wireless correspond-ence in a small space,”
in Knowledge-Based Intelligent Inform-ation and Engineering Sys-
tems. Berlin, Germany: Springer, 2004, pp. 898–904.
[42] S. Dawson-Haggerty, A. Tavakoli, and D. Culler, “Hydro: A hybrid
routing protocol for low-power and lossy networks,” in Proc. IEEE
Int. Conf. Smart Grid Commun., Oct. 2010, pp. 268–273.
[43] T. Iwao, K. Yamada, M. Yura, Y. Nakaya, A. Ćardenas, S. Lee, and R.
Masuoka, “Dynamic data forwarding in wireless mesh networks,” in
Proc. IEEE Int. Conf. Smart Grid Commun., Oct. 2010, pp. 385–390.
[44] J.-S.Jung,K.-W.Lim,J.-B.Kim,Y.-B.Ko,Y.Kim,andS.-Y.Lee,
“Improving IEEE 802.11s wireless mesh networks for reliable routing
in the smart grid infrastructure,” in Proc. IEEE Int. Conf. Commun.
Workshops, Jun. 2011, pp. 1–5.
[45] J. Kim, D. Kim, K. Lim, Y. Ko, and S. Lee, “Improving the reliability
of IEEE 802.11s based wireless mesh networks for smart gridsystems,”
J. Commun. Netw., vol. 14, no. 6, pp. 629–639, Dec. 2012.
[46] Q. Zhu, D. Wei, and T. Basar, “Secure routing in smart grids,” in Proc.
Workshop Foundations Dependable and Secure Cyber-Physical Syst.,
2011, pp. 55–59.
[47] B. Lichtensteiger, B. Bjelajac, C. Muller, and C. Wietfeld, “RF mesh
systems for smart metering: System architecture and performance,” in
Proc. IEEE Int. Conf. Smart Grid Commun., Oct. 2010, pp. 379–384.
[48] T. Zhu, S. Xiao, Y. Ping, D. Towsley, and W. Gong, “A secure energy
routing mechanism for sharing renewable energy in smart microgrid,”
in Proc. IEEE Int. Conf. Smart Grid Commun., Oct. 2011, pp. 143–148.
[49] G. Bumiller, T. Sauter, G. Pratl, and A. Treytl, “Secure and reliable
wide area power-line communication for soft-real-time applications
within REMPLI,” in Proc. IEEE Int. Symp. Powerline Commun. Ap-
plications, Apr. 2005, pp. 57–60.
[50] S. Rogai, “Telegestore project—Progress and results,” in Proc. IEEE
Int. Symp. Power Line Commun. Applications, Mar. 2007, pp. 1–1.
[51] M. Sebeck and G. Bumiller, “A network management system for
power line communications and its verification by simulation,” in
Proc. 4th Int. Symp. Power-Line Commun. Applications, Apr. 2000,
pp. 225–232.
[52] R. Brito, G. Bumiller, and Y. Song, “Modeling and simulation of a
SFN based PLC network,” in Proc. Int. Symp. Power Line Commun.
Applications, Apr. 2005, pp. 331–335.
[53] L. Lampe, R. Schober, and S. Yiu, “Multihop transmission in power
line communication networks: Analysis and distributed space-time
coding,” in Proc. IEEE Workshop Signal Process. Advances Wireless
Commun., Jun. 2005, pp. 1006–1012.
[54] G. Bumiller, L. Lu, and Y. Song, “Analytic performance comparison
of routing protocols in master-slave PLC networks,” in Proc. IEEE Int.
Symp. Power Line Commun. Applications, Apr. 2005, pp. 116–120.
[55] M. Biagi and L. Lampe, “Location assisted routing techniques for pow-
erline communication in smart grids,” in Proc. IEEE Int. Conf. Smart
Grid Commun., Gaithersburg, MD, USA, Oct. 2010, pp. 274–278.
[56] L. Lampe and A. Vinck, “Cooperative multihop power line communi-
cations,” in Proc. IEEE Int. Symp. Power Line Commun. Applications,
Mar. 2012, pp. 1–6.
[57] H. Gharavi and B. Hu, “Multigate communication network for smart
grid,” Proc. IEEE, vol. 99, no. 6, pp. 1028–1045, Jun. 2011.
[58] D. Dietrich, D. Bruckner, G. Zucker, and P. Palensky, “Com-
munication and computation in buildings: A short introduction
and overview,” IEEE Trans. Ind. Electron., vol. 57, no. 11, pp.
3577–3584, Nov. 2010.
[59] C. Zimmer and F. Mueller, “Fault-tolerant network routing through
software overlays for intelligent power grids,” in Proc. IEEE Int. Conf.
Parallel Distrib. Syst., Dec. 2010, pp. 542–549.
[60] K. Gill, S. Yang, F. Yao, and X. Lu, “A ZigBee-based home automation
system,” IEEE Trans. Consum. Electron., vol. 55, no. 2, pp. 422–430,
May 2009.
[61] V. C. Gungor, B. Lu, and G. P. Hancke, “Opportunities and challenges
of wireless sensor networks in smart grid,” IEEE Trans. Ind. Electron.,
vol. 57, no. 10, pp. 3557–3564, Oct. 2010.
[62] C.-W. Lu, S.-C. Li, and Q. Wu, “Interconnecting ZigBee and
6LoWPAN wireless sensor networks for smart grid applications,” in
Proc. IEEE Int. Conf. Sensing Technol., Dec. 2011, pp. 267–272.
[63] H. Hrasnica, A. Haidine, and R. Lehnert, Broadband Powerline Com-
munications Networks: Network Design. Hoboken, NJ, USA: Wiley,
2004.
[64] S. Güzelgöz, H. Arslan, A. Islam, and A. Domijan, “A review of wire-
less and PLC propagation channel characteristics for smart grid envi-
ronments,” in Journal of Electrical and Computer Engineering,Jan.
2011, pp. 1–12.
[65] S. Galli, A. Scaglione, and Z. Wang, “Power line communications and
the smart grid,” in Proc. IEEE Int. Conf. Smart Grid Commun.,Oct.
2010, pp. 303–308.

SABBAH et al.: SURVEY OF NETWORKING CHALLENGES AND ROUTING PROTOCOLS IN SMART GRIDS 221
[66] R. Lewis, “Assessment of communication methods for smart electricity
metering in the U.K.,” in Proc. IEEE Conf. Sustainable Alternative
Energy, Sep. 2009, pp. 1–4.
[67] S. Ullo, A. Vaccaro, and G. Velotto, “The role of pervasive and coop-
erative sensor networks in smart grids communication,” in Proc. IEEE
Mediterranean Electrotech. Conf., Apr. 2010, pp. 443–447.
[68] J. Gao, Y. Xiao, J. Liu, W. Liang, and C. Chen, “A survey of com-
munication/networking in smart grids,” J. Future Generation Comput.
Syst., vol. 28, no. 2, pp. 391–404, Feb. 2010.
[69] W. Wang, Y. Xu, and M. Khanna, “A survey on the communication
architectures in smart grid,” Comput. Networks, vol. 55, no. 15, pp.
3604–3629, Oct. 2011.
[70] X. Fang, S. Misra, G. Xue, and D. Yang, “Smart grid—The new and
improved power grid: A survey,” IEEE Commun. Surveys Tutorials,
vol. 14, no. 4, pp. 944–980, 4th Quarter 2012.
[71] T. Sauter and M. Lobashov, “End-to-end communication architecture
for smart grids,” IEEE Trans. Ind. Electron., vol. 58, no. 4, pp.
1218–1228, Apr. 2011.
[72] “PRIME: PHY, MAC, and Convergence Layers,” PRIME Project,
2008.
[73] D. Gaorfalo, “Smart Metering and Smart Grids: The Enel Experience,”
[Online]. Available: http://www.ceps.eu/system/files/events/2010/02/
Garoflo_ENEL.pdf
[74] “The Evora Project,” [Online]. Available: http://www.inovcity.pt/pt/
Pages/homepage.aspx
[75] “Mitsubishi Electric,” [Online]. Available: http://www.mitsubishi-
electric.com/company/environment/report/products/randd/smart-
grid/index.html
[76] “General Motors Smart Grid Pilot,” [Online]. Available: http://
media.gm.com/content/media/us/en/gm/news.detail.html/con-
tent/Pages/news/us/en/2011/Jul/0719_onstar.html
[77] M. Ibnkahla, Wireless Sensor Networks: A Cognitive Perspective.
Boca Raton, FL, USA: CRC Press—Taylor and Francis, 2012.
[78] A. El-Mougy and M. Ibnkahla, “A cross-layer framework for network
management in wireless sensor networks using weighted cognitive
maps,” Int. J. Distrib. Distributed Sensor Netw., vol. 2013, pp. 1–18,
2013.
Ayman I. Sabbah (S’04) received the B.Sc. degree
in communications and control engineering (with
distinction) from University of Gaza, Palestine, in
2007, and the M.S. degree in wireless communica-
tions engineering (summa cum laude) from Jordan
University of Science and Technology, Jordan, in
2010. He is currently working toward the Ph.D. de-
gree at Queen’s University, Kingston, ON, Canada.
He is a Research Assistant with Queen’s Univer-
sity, Kingston, ON, Canada. He won DAAD schol-
arship and Hani Qaddumi Scholarship. His research
interests are cognitive radio for WSNs, dynamic spectrum allocation, and smart
grid communications.
Amr El-Mougy received the M.Sc. degree from
Concordia University, Montreal, QC, Canada, in
2006, and the Ph.D. degree from Queen’s University,
Kingston, ON, Canada, in 2013.
He currently holds a postdoctoral position with the
University of Ottawa, Ottawa, ON, Canada, working
mainly on wireless broadband networks for public
safety. He has authored and coauthored more than a
dozen journal and conference publications as well as
two book chapters. He has also cosupervised several
graduate and undergraduate students. In addition, he
was also the coordinator for project titled “Versatile Wireless Sensor Network
for Environmental Monitoring”, which involved the design and implementation
of a WSN. He was also a member of the teams working on projects that involved
a WSN for highway safety and a WSN for smart grids.
Mohamed Ibnkahla (M’92) received the Ph.D.
degree and Habilitation a Diriger des Recherches de-
gree (HDR) from the National Polytechnic Institute
of Toulouse (INP), Toulouse, France, in 1996 and
1998, respectively.
In 2000, he joined the Department of Electrical
and Computer Engineering, Queen’s University,
Kingston, ON, Canada where he is now a Full
Professor. He is currently involved in a number
of projects applying wireless sensor technology
to various areas such as smart grid communica-
tions, environment monitoring, wildlife tracking, precision agriculture, food
traceability, highway safety, intelligent transportation systems and water
management. He has published more than 120 technical papers and book
chapters. He has authored or edited five books: Signal Processing for Mobile
Communications Handbook (Taylor and Francis—CRC Press, 2004), Adaptive
Signal Processing in Wireless Communications (Taylor and Francis—CRC
Press, 2008), Adaptive Networking and Cross-layer Design in Wireless Net-
works (Taylor and Francis—CRC Press, 2008), Wireless Sensor Networks: A
Cognitive Perspective (Taylor and Francis—CRC Press, 2012), and Cooper-
ative Cognitive Radio Networks: The Complete Spectrum Cycle (Taylor and
Francis—CRC Press, to be published in 2014).