2327-4662 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2593438, IEEE Internet of
Things Journal
1
An Energy-efficient Region-based RPL Routing
Protocol for Low-Power and Lossy Networks
Ming Zhao, Student Member, IEEE, Ivan Wang-Hei Ho and Peter Han Joo Chong, Member, IEEE
Abstract—Routing plays an important role in the overall archi-
tecture of the Internet of Things. IETF has standardized the RPL
routing protocol to provide the interoperability for Low-Power
and Lossy Networks (LLNs). LLNs cover a wide scope of applica-
tions, such as building automation, industrial control, healthcare
and so on. LLNs applications require reliable and energy-
efficient routing support. Point-to-point (P2P) communication is a
fundamental requirement of many LLNs applications. However,
traditional routing protocols usually propagate throughout the
whole network to discover a reliable P2P route, which requires
large amount energy consumption. Again, it is challenging to
achieve both reliability and energy-efficiency simultaneously,
especially for LLNs. In this paper, we propose a novel energy-
efficient region-based routing protocol, called ER-RPL, which
achieves energy-efficient data delivery without compromising
reliability. In contrast of traditional routing protocols where all
nodes are required for route discovery, the proposed scheme only
requires a subset of nodes to do the job, which is the key of energy
saving. Our theoretical analysis and extensive simulation studies
demonstrate that ER-RPL has a great performance superiority
over two conventional benchmark protocols, i.e., RPL and P2P-
RPL.
Index Terms—Low-power and Lossy Networks (LLNs), RPL,
region-based, energy-efficiency, reliability, Point-to-point (P2P)
communication.
I. INTRODUCTION
M
ACHINE-to-Machine (M2M) communications [1] aim
to achieve ubiquitous communication among intelli-
gent devices for application control and monitoring, which
have attracted both academia and industry in recent years.
Motivated by the great potential of M2M communications,
many standardization activities, such as IETF, 3GPP, and
IEEE have defined protocol stacks to enable M2M commu-
nications [2]. M2M devices are usually battery powered and
operate in harsh environment (e.g., heat, dust, and moisture
weather). The dynamic and lossy environment where a large
group of highly constrained devices are interconnected by
unreliable wireless links are categorized as Low-power and
Lossy Networks (LLNs). Routing plays a crucial role to
M. Zhao is with the School of Electrical and Electronic Engineering,
Nanyang Technological University, Singapore.
E-mail: zhaoming@ntu.edu.sg
Ivan W. H. Ho is with the Department of Electronic and Information
Engineering, The Hong Kong Polytechnic University, Hong Kong.
E-mail: ivanwh.ho@polyu.edu.hk
P. H. J. Chong is with Department of Electrical and Electronic Engineering,
Auckland University of Technology, New Zealand.
E-mail: peter.chong@aut.ac.nz
Copyright (c) 2012 IEEE. Personal use of this material is permitted.
However, permission to use this material for any other purposes must be
obtained from the IEEE by sending a request to pubs-permissions@ieee.org
provide the interoperability among network components. The
IETF Routing Over Low-power and Lossy network (ROLL)
working group standardizes the IPv6 Routing Protocol for
Low Power and Lossy Networks (RPL) [3], [4], [5], which
targets highly constrained nodes and large scale networks for
LLNs applications. Point-to-point (P2P) communication is a
fundamental requirement of most of the LLNs applications.
LLNs applications require efficient P2P routing support.
However, achieving high reliability and consuming less energy
at the same time are inherently challenging. The routing
paths between arbitrary M2M devices are not provided by
default due to the resource constraints. Traditional routing
protocols, such as Lightweight On-demand Ad hoc Distance-
vector Routing (LOADng) [6] and P2P-RPL [7], dissemi-
nate route discovery messages throughout a network to find
the optimal P2P routes, leading to large amount of routing
overhead and energy consumption. In this paper, we propose
an energy-efficient region-based routing protocol (ER-RPL),
which achieves energy-efficient P2P data delivery without
compromising the reliability. In contrast to traditional routing
protocols, in which all nodes need to participate in the route
discovery, ER-RPL only requires a subset of nodes in some
regions to discovery the route. In ER-RPL, a nearly optimal
route in terms of reliability can be discovered with a great
energy conservation.
Our proposed ER-RPL makes use of the region information
to support efficient P2P communication. For static networks,
such as M2M networks and wireless sensor networks (WSNs),
the area where a node resides is a piece of important informa-
tion. Many LLNs applications exploit this region feature. For
example, in automatic control systems, the control command
can be sent to all devices in one level or a region/room of a
building. In event-triggered applications, all sensors within an
area can capture this event, but which sensor has collected the
information may not be important. The region information can
be used to efficiently discover the routing paths. Overall, the
key contributions of this paper are summarized as follows:
1) We propose a novel scalable routing protocol, i.e., ER-
RPL, to achieve reliable and energy-efficient data deliv-
ery for static networks. Significant reduction on control
overhead is achieved, since only a portion of nodes in
the network need to participate in the route discovery.
2) We propose a P2P traffic model with the consideration
of routing decision for lossy networks.
3) Both theoretical analysis and simulation studies are
performed to evaluate the effectiveness and flexibility
of our proposed ER-RPL.
4) Two conventional routing protocols, RPL and P2P-RPL,
2327-4662 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2593438, IEEE Internet of
Things Journal
2
are used as the benchmarks in the simulation study. In
comparison with the benchmarks, ER-RPL can achieve
more reliable data delivery with significant energy con-
servation.
The rest of this paper is organized as follows. Section II
introduces existing LLNs routing protocols and discusses the
challenges of the design for efficient LLNs routing protocol.
Section III presents the proposed protocol ER-RPL. Theoreti-
cal analysis are presented in Section IV. Extensive simulation
and performance evaluation are conducted in Section V. Fi-
nally, this paper is concluded in Section VI.
II. BACKGROUND
In recent years, LLNs applications, which cover a wide
range of scenarios, including but not limited to, building
automation, industrial control, urban environment and home
automation [8], [9], [10], have emerged as the predominant
paradigm for M2M communications. LLNs applications essen-
tially require reliable and energy-efficient routing to support
the connectivity of network utilities with tighter control and
energy conservation. However, nodes operating in LLNs usu-
ally have limited battery power and communicate via dynamic
and lossy wireless medium. It is inherently challenging to
achieve reliability and energy-efficiency at the same time,
especially for LLNs. It has been shown that traditional routing
protocols face difficulties in providing an efficient routing
support for LLNs [11]. For P2P communication, three main
routing techniques have been proposed: proactive routing, on-
demand (reactive) routing and geographic routing. In this
section, we first introduce several conventional LLNs routing
protocols and analysis their limitations. Then we summarize
the main challenges of designing routing protocols for LLNs,
and present the routing solutions that are implemented in our
proposed ER-RPL.
A. Routing Protocols for Low-Power and Lossy Networks
RPL is a proactive IPv6-based distance-vector routing proto-
col. RPL can establish a Destination Oriented Directed Acyclic
Graph (DODAG) at a high speed with the trickle algorithm
[12]. According to applications’ objectives, RPL uses different
routing metrics to support LLNs applications. The root node
serves as a transit point to bridge the DODAG with the IPv6
network. The formation of a DODAG is initiated by the root
node that periodically originates DODAG Information Object
(DIO). RPL is designed to optimize the routing support for
multipoint-to-point (MP2P). RPL chooses the best next hop
as the preferred parent to the root node given a particular
objective function [13]. Although RPL can support the routing
for generic traffic pattern, RPL needs to pre-establish routes
and can only route along pre-established DAGs for P2P com-
munication. The source node has to send the packet upwards
until it reaches the ancestor node of the destination node. Then
the common ancestor node delivers the packet downwards
towards the destination node. In non-storing mode of RPL,
the common ancestor has to be the root node. Hence, packets
need to travel through many lossy links, resulting in long end-
to-end delay. Additionally, the root becomes a bottleneck when
traffic load becomes heavy.
c
a
b
d
e
f g
i
j
lk
Root
h
Fig. 1 Communication routes between node h and node f, provided by RPL with the
dotted arrows and by P2P-RPL with solid arrows.
A reactive P2P route discovery mechanism based on RPL
is defined as P2P-RPL [7], [14]. In P2P-RPL, a temporary
DAG, in which the source node acts as the root node, is built
to facilitate the end-to-end traffic transmission in LLNs. The
lifetime of the DODAG is strictly restricted by the lifetime of
the route request. A P2P Route Discovery Option (P2P-RDO),
which is piggybacked in DIO, is used for the route discovery in
P2P-RPL. The source node originates and disseminates route
discovery messages throughout the whole network. The fre-
quency of broadcasting route discovery messages is according
to the trickle algorithm. Once the destination node receives
the P2P-RDO, it replies a P2P Discovery Reply Object (P2P-
DRO) to the source through the discovered route. The reverse
route of P2P-DRO is used for P2P data delivery. Fig. 1 shows
the communication routes for node h to node f in RPL and
P2P-RPL, respectively. P2P-RPL can usually find better P2P
route than RPL, but sometimes the route discovered by P2P-
RPL may not be much better than the existing route in RPL.
The improvement of route quality depends on the network
topology, such as the distance between the source node and
destination node. Moreover, the cost associated with the P2P
route discovery is very costly in terms of energy consumption,
especially for LLNs. Because all nodes in the network need to
participate in the formation of temporary DODAGs during the
route discovery, resulting in significant energy consumption
for energy constrained networks.
LOADng is a reactive routing protocol derived from Ad
hoc On-demand Distance Vector Routing (AODV) [15] for
LLNs. The operation of LOADng makes lots of simplifications
on AODV. During the route discovery stage, route request
(RREQ) messages will be distributed throughout a network.
If the destination node receives a RREQ message, it replies a
route-reply (RREP) message through the reverse path to the
source node. LOADng supports many traffic patterns, such
as P2P, point-to-multipoint (P2MP), and MP2P. However, for
MP2P traffic pattern, the routing overhead in LOADng is much
more than that in RPL [16]. The performance comparison and
detail analysis of LOADng and RPL have been conducted
with different traffic patterns [17], [18]. Both P2P-RPL and
LOADng disseminate route discovery messages throughout the
whole network. The route discovery scheme in LOADng is
similar to that in P2P-RPL, so LOADng is not addressed in
the simulation study.
2
2327-4662 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2593438, IEEE Internet of
Things Journal
3
Geographic routing relies on the locations of nodes (either
real coordinates or virtual coordinates) instead of nodes’ IP
addresses to forward data packets in a greedy manner [19],
[20]. A node chooses the node, which is the closest one to
the destination from its neighbors, as its next hop to relay
its data packets. The location/coordinate is obtained by each
node or partial nodes either as a priori knowledge or through
a self-configuration localization scheme. Geographic routing
has the advantage of low routing overhead and scalability
support, but it does not take into consideration of the lossy
nature of wireless links in the selection of the next hop. Con-
sequently, geographic routing usually cannot cope well with
the lossy wireless medium to provide reliable data delivery
support for LLNs. Additionally, in some geographic routing
protocols, nodes have to exchange the one-hop or even two
hop neighbour table periodically to maintain the coordinates
[20]. It is very costly in terms of energy consumption for a
resource constrained network.
B. Routing Challenges in Low-Power and Lossy Networks
Efficient support for generic traffic patterns. LLNs
applications require stable routing support for generic traffic
patterns, such as MP2P, P2P, and P2MP with heterogeneous
node capabilities [11]. However, devices in LLNs applications
have limited memory. In order to keep the routing table size
to be small, the routing path between two arbitrary nodes is
usually not provided by default. Thus, the route discovery is
required when there is no available route between a source
and destination pair.
Reliable routing in dynamic and lossy environment. It is
extremely crucial and challenging to achieve reliable routing
in dynamic and lossy environment. Data transmission suffers
from link loss in LLNs. In addition to harsh environment,
channel fading and co-channel interference add more uncer-
tainties to data transmission in wireless channels. Retransmis-
sions usually result from unreliable wireless channels, leading
to higher energy consumptions and longer channel occupancy
time. It is vital to discovery reliable routing paths for data
delivery in LLNs.
Energy-efficient route discovery. In LLNs, a large number
of energy constrained nodes may be potentially inaccessible
due to the constraints of physical environment in realistic.
Conserving power and prolonging the lifetime of the network
are critical to maintain persistent network connectivity so as to
attain good network performances. The expenditure of energy
results from the transmission and reception of data packets
and control packets. Control packets are commonly used for
topology construction in proactive routing protocols and route
discovery in reactive routing protocols. For nodes with severe
energy constraints, the transmission and reception of control
packets are very costly in terms of energy consumption. In
addition, nodes in LLNs are usually battery-powered and have
to run complex computational processes. Traditional routing
protocols disseminate route discovery messages throughout the
whole network in all directions, which incurs very much over-
head. Consequently, routing overhead should be minimized so
as to conserve energy in the design of routing protocols for
LLNs.
Link asymmetry in real-world scenarios. Empirical stud-
ies have shown that wireless links have asymmetric nature.
The main reason is that the transmitter power and receiver
sensitivity are different from nodes to nodes [22]. Some other
factors, such as the reflectors, absorbers, etc, also result in
link asymmetry. The asymmetric nature of wireless links has
significant impact on the routing protocols’ performance, espe-
cially for LLNs. Protocols without considering the asymmetry
of wireless links sometimes fail when it is encountered in the
real deployments [23]. Therefore, the asymmetry properties of
wireless channels needs to be considered in the design.
Scalability support for large scale networks. LLNs are
normally large scale and require routing protocols to provide
scalability support. Routing protocols for LLNs have to be
scalable so as to support large and increasing number of nodes.
Scalability is one of the most important criteria in the design
of LLNs routing protocols [21].
To address the above challenges and limitations of existing
LLNs routing protocols, our proposed ER-RPL consists of the
following four major components:
1) ER-RPL is designed with the capability to support
generic traffic patterns, because it takes advantage of
the existing DODAG structure of RPL in addition to its
efficient support for P2P route discovery.
2) ER-RPL exploits the region feature of static networks.
Only a portion of nodes are required to participate
in the route discovery for a source-destination node
pair. In addition to the region-based route discovery,
Region-to-Region (R2R) routing without route discovery
is implemented as an enhancement. These designs result
in a great reduction of routing overhead. In this way,
reliable routing paths can be discovered in an energy-
efficient manner.
3) A distributed Self-regioning algorithm is proposed for
nodes to compute their region codes (RCs). Meanwhile,
the region-based route discovery also works in a decen-
tralized manner. Therefore, ER-RPL has the capability
to support network scalability.
4) The asymmetric nature of wireless links are considered
in the protocol design. ER-RPL is robust to different
wireless channel conditions.
III. PROPOSED PROTOCOL: ER-RPL
Based on the key challenges and insights described in the
Section II, we propose a hybrid of proactive and reactive
routing protocol, namely ER-RPL, to achieve reliable data
delivery in an energy-efficient manner. In this section, we first
provide the system model of ER-RPL, and then present the
key stages in ER-RPL for efficient P2P route discovery and
data delivery.
A. System Model
1) Preliminary: We consider that n stationary nodes are
densely deployed in the area with the size of G (m
2
).
Nodes with mobility are out of the scope of this paper.
The transmission range of nodes is R. In the network, we
consider a set of nodes with location-awareness capability
3
2327-4662 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2593438, IEEE Internet of
Things Journal
4
A
B
C
D
Reference Node Normal Node
Fig. 2 An example of reference node and the regions in the network
(e.g., with GPS), which are called Reference Nodes (RNs).
Fig. 2 shows an example of the RNs that are distributed in
the network. Assume that N denotes the number of RNs,
where N n. The set of all RNs in the network is denoted
by Ω = {RN
1
, RN
2
, . . . , RN
N
}. With the help of RNs,
the network area are segmented into several regions. The
nodes without position knowledge are regarded as normal
nodes (“nodes” in the following of this paper). The objective
of protocol design is to energy-efficiently discover reliable
routes for nodes without location-awareness capability so as
to support reliable and energy-efficient P2P communication.
Remote control applications typically require P2P communi-
cation. For example, a motion sensor (node s) suddenly needs
to communicates with a lamp module (node d). Node s and
node d are normal nodes in the network and do not have the
location-awareness capability. The source node s needs to send
the control command to the destination node d via multihop
routing. In ER-RPL, a reliable route between node s and node
d can be discovered in an energy-efficient manner.
Our proposed ER-RPL makes use of a small amount of
location-aware RNs, but it differs from geographic routing in
three fundamental ways: ER-RPL establishes the best quality
route in terms of reliability in a reactive manner. 2) In ER-
RPL, nodes do not have the unique geographic coordinates
knowledge. The region information is only used for nodes to
determine the necessity to participate in the route discovery.
3) In ER-RPL, the data delivery relays on the node’s address
and routing table instead of virtual or real coordinates. These
features make our proposed ER-RPL essentially different from
geographic routing protocols.
Our study considers two scenarios. In the first scenario, the
node are nearly uniformly distributed so that the value of ρ is
a constant. In the second scenario, the value of ρ is likely to
change due to the irregular node deployment. Table I shows
some notations to be used in this paper.
2) P2P Traffic Modelling in LLNs: Packets drop takes
place in both node level and link level [22]. The link level
channel contention and node level resource limitation affect
communication quality. In this research study, we assume that
Table I Summary of notations
Symbol Definition
n number of nodes
N number of reference nodes
G network size
Ω The set of all RNs
ρ network density
R the transmission range of node and RN
λ the number of traffic flows
I
max
the maximum time interval size
p
ij
the successful delivery probability from node i to j
E
ij
one hop ETX from node i to j
C
I
the Coordinate of RN I
h
iI
the hop count of arbitrary node i to RN I
RNM(i, j) or
RN
ij
the entry in the ith row and jth column of the matrix
RNM
RCM (m, n)
or RC
mn
the entry in the mth row and mth column of the
matrix RCM
nodes have enough buffer space, thus packets drop due to
buffer overflow are negligible in this study.
Assume that the link loss is independent and identically
distributed (i.i.d). The successful delivery probability from
node i to node j can be denoted by p
ij
. We assume that wireless
links have bidirectional readability and can be asymmetric,
which means p
ij
may not be equal to p
ji
for the delivery of
packets. The Medium Access Control (MAC) layer has the
retransmission scheme implemented to improve the reliability
of transmission. The Expected Transmission count (ETX) [24]
is used to measure the quality of wireless links. One hop
ETX means the average number of transmissions required
to successfully deliver one packet to the next hop, which is
defined as E
ij
= 1/p
ij
p
ji
, where p
ij
is the probability of
successful delivery of a data packet from node i to i, and p
ji
in this equation refers to the probability of successful delivery
of an ACK from node j to node i. Because the ACK usually
has very small size, which can be recovered with the strong
coding techniques. Hence, p
ji
for ACK is approximate to one.
In this study, we assume ACK does not suffer from link loss,
so we can get E
ij
= 1/p
ij
. The aggregate ETX is used to
select the best route for a source-destination node pair.
The traffic load from source node i to destination node j is
denoted by L
ij
, where i 6= j. Then the traffic load distribution
matrix of the network is
L =
L
00
··· L
0N
.
.
.
.
.
.
.
.
.
L
N0
··· L
NN
, (1)
where for i ∈ n, L
ii
= 0.
For P2P communication, a node needs to choose the next
hop among its neighbouring nodes for packet forwarding.
This routing decision has significant impact on the network
performance. To elucidate this, let δ
ij
be the routing choice
made by node i to node j, where j ∈ N(i) and N(i) denotes
node i’s neighbour node set. If δ
ij
is equal to 1, it means that
node i chooses node j as its next hop. If the value of δ
ij
is 0,
it means that node i does not choose node j as its next hop.
We use r
i,s
to denote the traffic generation rate at node i, and
4
2327-4662 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JIOT.2016.2593438, IEEE Internet of
Things Journal
5
use r
i,d
to denote the arrival rate of the traffic destinated at
node i. Assume that the traffic receiving rate of node i from
node j is f
j,i
, and the traffic departure rate at node i to node
j is f
i,j
. In this way, the P2P traffic model for node i can be
formulated as
r
i,s
+
X
j∈N(i)
δ
ji
f
ji
−
X
j∈N(i)
δ
ij
f
ij
p
ij
= r
i,d
, (2)
where 0 ≤ f
ji
≤ C
ji
, ∀i, j ∈ n and i 6= j.
Let us define the network running time as T. At time t, the
traffic generation rate of node i is r
i,s,t
. The traffic arrival rate
of node i as destination is represented by r
i,d,t
. The duration
of traffic flow is ∆t. In this way, the traffic load of node i is
shown in Eq. (3) and (4).
Z
T
0
r
i,s,t
∆t
i
= L
i0
+ L
i1
+ ... + L
iN
=
X
0≤j≤N
L
ij
, (3)
Z
T
0
r
i,d,t
∆t
i
= L
01
+ L
1i
+ ... + L
Ni
=
X
0≤j≤N
L
ji
. (4)
3) Energy Model: We model nodes with four basic states
(transmit, receive, idle, sleep) and a transition state among
them [25]. The energy consumption in transmitting and receiv-
ing states are the major components to be considered in this
study. The energy consumption is modelled using the First Or-
der Radio Model [26], which has been widely used to measure
the energy dissipation for wireless sensor networks (WSNs)
[27], [28]. In this model, the radio consumes E
elec
to run the
transmitter or receiver circuitry. The transmitting amplifier is
amp
, which is used to achieve the acceptable signal-to-noise
(SNR) ratio. We assume the propagation loss exponent is 2.
In this way, the energy consumption for transmitting a l-bit
message with a transmission range R is modelled as
E
tx
(l, R) = lE
elec
+ lR
2
amp
, (5)
The energy consumption of the receiver is modeled as
E
rx
(l, d) = lE
elec
. (6)
B. Overview of ER-RPL
ER-RPL inherits the mechanism from RPL to pre-establish
the DODAG such that it can support the multipoint-to-point
(MP2P) with the optimized topology. Additionally, ER-RPL
discovers the best P2P route with the region information in
a reactive manner. Both RNs and nodes are stationary in this
work. In ER-RPL, each RN belongs to an area, and that area is
divided into a configurable number of non overlapping regions.
The region number associated with each RN can be configured
with the Self-regioning algorithm. Then each node starts to
estimate which region it resides in. ER-RPL requires a node
to determine the necessity to participate in the route discovery
based on the knowledge of the source and destination nodes’
region. In this way, the route discovery is only performed
among a subset of nodes in the network, leading to significant
reduction of routing overhead. Basically, ER-RPL includes two
main stages. 1). Network initialization stage: Each RN com-
putes its Coordinates, which is defined as the average distance
per hop count. Then, each node estimates the distances to
each RNs based on the RN’s Coordinates values and their
hop counts to the corresponding RN. The particular region
that a node resides in the network is represented by a region
code (RC). With the distributed Self-regioning algorithm, a
node can compute its RC in ER-RPL. 2). Route discovery
stage: According to the RC of source node and destination
node, the route discovery is only performed among a subset
of nodes in the network with the region-based route discovery.
Besides, the Region-to-Region (R2R) routing without route
discovery is designed as an enhancement to the region-based
route discovery.
C. Network Initialization Stage
A series of positioning algorithms have been proposed,
which are well known as the Ad Hoc Positioning Sys-
tem (APS). In particular, APS includes six algorithms: DV-
Distance, DV-Hop, Euclidean, DV-Bearing, DV-Coordinate,
and DV-Radial [29]. Different from the work in [29], where the
Coordinates is defined as a correction factor, our work defines
the Coordinates as the average distance per hop of RNs. ER-
RPL is designed with an enhanced DV-Hop algorithm, which
uses RNs’ Coordinates and the hop counts to RNs for the
estimation of the distances between a node to the RNs. It is
worthy to highlight that the number of RNs in a network is
known in advance by all RN nodes. RNs play a critical role in
the network initialization stage, but they do not perform any
task during the reactive P2P route discovery and data delivery.
1) Coordinates of Reference Node: Each RN computes its
Coordinates value based on its relative distances and hop
counts towards other RNs in the network. Upon receiving the
topology formation information initiated by the root node, each
RN serves as a temporary “root” and builds up a temporary
DODAG using the Minimum Hop Count [30] as the routing
metric. The construction of temporary DODAG is achieved by
disseminating Region Formation Objec (RFO) messages, and
different regions are formed during this process. Fig. 3 shows
the packet structure of a RFO message. The DODAG
RN is
the IP address of the RN. The Rank shows the hop distance
to that RN. The Position carries the geographical position of
this RN. There are two stages for the exchange of RFOs,
which are indicated by the CFlag. When the CFlag is set
to be zero by a RN, it indicates that the current stage is the
Coordinates Computation stage for this RN. When the CFlag
is set to be one, it is in the Euclidean Distance Calculation
stage. The Coordinate field records the Coordinates value of
the corresponding RN. The temporary DODAG rooted at RN
I is denoted as DODAG RN
I
, where I is an arbitrary RN in
the network area. Upon receiving RFO messages, a node first
checks the CFlag field to determine the stage of the RN .
During the Coordinates Computation stage, a node joins
the temporary DODAG, chooses the next hop, updates its
rank, and broadcasts its status with RFO messages. Similar to
DODAG_RN
Rank CFlag Coordinate Position
Fig. 3 The Packet Structure of RFO Message
5
