Está en la página 1de 10

Peer-to-Peer Netw. Appl.

DOI 10.1007/s12083-015-0349-8

OPNET-based modeling and simulation of mobile Zigbee


sensor networks
Xiaolong Li1 · Meiping Peng2 · Jun Cai3 · Changyan Yi3 · Hong Zhang3

Received: 7 January 2015 / Accepted: 23 March 2015


© Springer Science+Business Media New York 2015

Abstract Modeling and simulation can help to validate layer process model and the implementation procedures of
and evaluate the performance of wireless sensor networks its kernel functions. Comprehensive performance compar-
(WSNs) within specific applications. In order to resolve the isons are performed between the proposed model and the
issue of the restriction on node mobility in existing Zig- Zigbee model in OPNET standard libraries. In order to eval-
bee WSN simulation models, this paper proposes a Zigbee uate the effectiveness of the proposed model in the aspect
compliant new simulation model using the OPNET simu- of node mobility support, time intervals between route fail-
lator. Based on the Zigbee MAC layer model in OPNET ure occurrence and route recovery are measured as well.
Modeler, we develop a network layer model and propose an The experimental results show that the proposed simulation
improved AODV routing algorithm to support node mobil- model achieves better performance, compared to the orig-
ity, both of which are compatible with Zigbee protocols. inal one. In addition, when node mobility causes routing
We further present in details the structure of the network failures, alternative routes can be established quickly by the
proposed model.

Keywords OPNET · Zigbee · IEEE802.15.4 · Mobility ·


 Xiaolong Li
xlli@guet.edu.cn Simulation · Network layer

Meiping Peng
meipingp@hotmail.com 1 Introduction
Jun Cai
jcai@umanitoba.ca
Wireless sensor networks generally comprise a large num-
ber of sensor nodes deployed in an area of interest to collect
Changyan Yi physical or environmental conditions, such as temperature,
changyan.yi@umanitoba.ca. humidity, pressure, etc. In wireless sensor networks, per-
formance evaluation is critical to test the practicability of
Hong Zhang network architectures and protocol algorithms, and provides
hzhang@cc.umanitoba.ca
guidelines in performance optimization. Among different
candidates, simulation offers a cost-effective way. Recently,
1 Guangxi Key Laboratory of Trusted Software, researchers have developed many simulation models on
Guilin University of Electronic Technology,
different simulation platforms, such as OPNET, NS-2,
Guilin, China
TOSSIM, EmStar, OMNeT++, J-Sim, ATEMU, and Avrora
2 School of Computer Science and Engineering, [1]. Compared with other simulators , OPNET is more
Guilin University of Electronic Technology, suitable to simulate behaviors of networks in the real
Guilin, China
world. OPNET Modeler, as a network simulator, provides
3 Department of Electrical and Computer Engineering, an industry-leading network technology development envi-
University of Manitoba, Winnipeg, Manitoba, Canada ronment [2]. It can be used to design and study network
Peer-to-Peer Netw. Appl.

modeling and simulation in applications, equipments, proto- nodes. This motivated us to develop a new simulation model
cols and network communications, and show flexibility and based on the OPNET simulator for mobile Zigbee sensor
intuition in designing practical systems. networks.
Recently, Zigbee technology has been widely adopted to The main contributions of this paper are summarized as
develop wireless sensor network applications [3] by forming follows. 1) We adopt the OPNET simulation development
a wireless mesh network with low rate, low power con- platform to design a mobile Zigbee sensor network simu-
sumption, and secure networking. In Zigbee protocol stack, lation model compatible with Zigbee protocols, where the
the physical layer and the MAC layer protocols have been physical layer and the MAC layer defined by IEEE 802.15.4
defined by IEEE802.15.4 standard [4]. Its network layer are employed. 2) We provide a node level design of mobile
built upon both lower layers should be designed to enable sensor nodes, present a process level model of its network
a mesh networking, support node joining or leaving, assign layer model and the detailed implementation procedure of
network addresses to devices, and perform routing. Zigbee the key functions. 3) In order to further decrease the com-
Alliance is working at providing a standardized base set munication overhead of nodes, an improved AODV routing
of solutions for sensor networks [5]. In this paper, a net- algorithm is also proposed, which demonstrates superior
work layer model is proposed for mobile sensor networks capability in supporting node mobility.
in order to accomplish all defined functions. The applica- The rest of this paper is organized as follows. In
tion layer aims at providing the services for an application Section 2, we discuss the design of network process model
program, consisting of application support sub-layer, appli- in details. In Section 3, we propose a new simulation
cation framework, and Zigbee device object. Since this layer model which enables mobile support for Zigbee devices.
is related to specific applications, and is not the main focus Section 4 presents our simulation results and demonstrates
of this paper,the design of the application layer is omitted experimental comparison between our proposed model and
here. ZMOMSL. Section 5 draws conclusions.
The simulation of Zigbee sensor networks within
OPNET simulator has been attracting interests from
researchers. There are many research works on simula- 2 The design of simulation system model
tion modelling and evaluation of sensor nodes in OPNET
[6, 7]. For example, Kucuk et al. [6] presented a detailed 2.1 Design of node model
implementation methodology for their proposed positioning
algorithm, called M-SSLE. Shrestha et al. [7] proposed a As shown in Fig. 1, a Zigbee node model within OPNET
simulation model for new networking nodes equipped with Modeler typically incorporates the physical layer, the MAC
multiple radio technologies. However, few works focused layer, the network layer and the application layer. The physi-
on the simulation model of mobile sensor networks in liter- cal layer comprises a transmitter module, a receiver module,
ature. Device mobility is inevitable and must be conciliated and a wireless pipeline model. The wireless pipeline model
[8, 9], where lack of the support for simulation on mobile can be configured to build a real radio environment. In
Zigbee sensor network is a major limitation in this field of the MAC layer, Carrier Sense Multiple Access with Col-
research, evaluation and development. lision Avoidance (CSMA/CA) protocol is used. For the
In [10], the adequacy of current provisions for deal- network layer, following services are provided: forming
ing with different mobility cases was assessed. Simulation a network, nodes joining and leaving a network, network
results demonstrated that the current model in OPNET stan- address assignment, neighbor discovery, and route mainte-
dard libraries is ineffective in dealing with nodal mobility. nance discovery. The application layer is responsible for
Since OPNET Modeler provides a comprehensive simu- producing and processing sensing data. In the rest of the
lation environment for modeling distributed systems and paper, we will focus on the design of the network layer
communication networks, many simulation studies for Zig- model for mobile Zigbee sensor networks.
bee sensor networks were performed in OPNET simulator
[11–16]. According to the performance studies using the 2.2 The design of network layer model
Zigbee model within OPNET Modeler standard libraries
(ZMOMSL), there are several disadvantages on this model. Three types of devices are defined in the Zigbee standard
For example, its address assignment mechanism may waste framework: coordinator, router, and end device. Coordina-
address space, the high communication overheads may tor is responsible for forming a new network, storing the key
reduce network lifetime, and the network joining strategy parameters of the network and connecting to other networks.
may result in significant traffic collisions and jams [17, 18]. There is always a single coordinator in a Zigbee network.
Among all these disadvantages, the most critical issue is that In Zigbee-based WSNs, sink node typically plays the role
the Zigbee model can not support the mobility of device of network coordinator. Router has the routing capability.
Peer-to-Peer Netw. Appl.

and Ae (d+1, m) is assigned to the m−th end device child at


depth d+1, they can be obtained by the following equations:

Ar (d + 1, m) = Aparent + Cskip(d) × Ri (2)

Ae (d + 1, m) = Aparent + Cskip(d) × Rm + i (3)

where Ri ∈ [0, Rm ], and Aparent represents the address of


the parent, i ∈ [0, (Cm − Rm )].
Since the tree address allocation mechanism can provide
a simple and reliable routing method for the entire Zigbee
network, it is employed in our proposed simulation model
to assign a network address to a device node (a router or an
end device) when it joins the network at the first time. In
order to build a model of the IEEE 802.15.4/Zigbee proto-
cols supporting node mobility, we propose a network layer
process based on OPNET simulation platform as shown in
Fig. 2. In the network process module, by moving into the
init state and forcedly traversing to state wait, all types of
Fig. 1 The developed node model
devices begin executing initialization procedure.

Specifically, it could allow other devices to join the net- 1) If the node is a coordinator, it will first setup
work as its child nodes, and route data packets. End device a network. By executing the transition function
has no routing capability, which relies only on its parents wpan execute scan(), the coordinator will scan all
(the coordinator or a router) to route data packets. Com- channels to select an unoccupied channel. After that,
pared with coordinator and router, end device has simpler it will invoke the setnetwork state’s Enter execs
hardware structure. wpan zigbee setnetwork() to choose a network ID
Each device node has a 16-bit short address for and a and configure other network parameters. Then it will
64-bit extended address in a Zigbee network. The 64-bit move into the active state to deal with network join-
extended address is set by manufactures, similar to the MAC ing and leaving requests, or routing messages. Among
address which is unique for each node. The 16-bit short all functions associated with the active state, func-
address is dynamically assigned to the node by its parent tion wpan handle mac pk() will be executed when
coordinator or router when the node joins the network and receiving data packets from the MAC layer, while
it is similar to the IP address in the Internet network. Zigbee function aodv rte rrep hello message send()
standard uses a distributed address allocation mechanism achieves the routing functionality. By periodically
for assigning address to the node when it joins the network. broadcasting hello data packets and receiving hello
Network address is determined by the following network packets from neighbors, the nodes (except the end
parameters which is provided by coordinator: *Cm the max- devices) will update their routing tables and neighbor
imum number of children allowed for each router, *Rm the tables. For end devices, they will choose one proper
maximum number of routers as subrouters linked to each node from candidate routers within its neighborhood
router, and *Lm the maximum depth of the whole network. as its parent.
The coordinator decides the depth of the whole network. 2) If the node is not the coordinator, it will scan all chan-
The size of the address sub-block allocated by each parent nels by executing function wpan execute scan(). If it
at depth d, Cskip(d) , could be described as: finds an available channel in which a network is oper-
ating, it will transit to J oin network state and begin
 to execute function wpan zigbee j oin network(),
1 + Cm (Lm − d − 1) Rm = 1
Cskip(d) = Lm −d−1
1+Cm −Rm −Cm Rm (1) where it will send a J OI NRequest packet to the rel-
1−Rm Rm > 1 evant router or coordinator asking to join the network
where Cskip(d) = 0 means that the node has no capability to and then wait for the response to its request. After
accept child nodes, and Cskip(d) > 0 otherwise. If network receiving the corresponding J oinResponse packet,
address Ar (d + 1, m) is assigned to the m − th router child the node will enter the active state.
Peer-to-Peer Netw. Appl.

2.3 Network join 0.2s is extremely adequate for awaiting and receiving reply
frame from its neighboring nodes, even considering the pro-
As shown in Fig. 2, all deployed nodes begin at the init cession delay, the transmission delay and the propagation
state. Each node is categorized as coordinator, router, or end delay. During this period of time, after receiving beacon
device. MAC layer channel scan is immediately called by frames, node A stores the neighbor information in its neigh-
the coordinator after the process of network initialization, bor table. At the end of the timer, the MAC layer schedules
and its status would be changed to set network. When the a remote interrupt to notice the network layer. Then, node
MAC layer channel scan is completed, coordinator will look A selects one node (coordinator or router) with the smallest
for an appropriate channel for establishing a new Zigbee hop count from itself as its potential parent node, denoted
network. After the suitable radio channel is found, the coor- by B, which has both the capability and permit to accept
dinator will assign a network identifier to the new network, new child nodes. Then, node A sends a join request frame to
which does not conflict with other existing networks. Then node B, and starts a timer to wait for the corresponding join
it will assign a network address to itself. Its status would be response frame. The initial value of the timer is set as 0.2s.
switched to the active state. After the coordinator finishes If it receives node B’s joining response frame before time-
the above operations, the network is formed. Since then, out, node A joins the network successfully. Its process status
other nodes will have the opportunity to join the network. moves into active state for data communications. Other-
The details of network join implementation procedure are wise, node A selects a new appropriate parent node in its
described below and are illustrated in Fig. 3. neighbor table and sends a new join request frame. How-
When a child node A wants to send the join request, A ever, if there is no appropriate parent node in the neighbor
first carries out the channel scan procedure at the MAC layer table, it will call for MAC layer channel scan again. The
and then broadcasts a beacon request frame. The process process status will stay in j oin network until node A joins
status is then transferred to j oin network. Meanwhile, a a network successfully.
timer of channel sensing duration is started. In our model, For a parent node, once it receives the beacon request
the initial value is chosen in the interval [0.2s, 0.4s], wherein frame, it will broadcast a new beacon frame. When it

Fig. 2 Zigbee network layer


model in OPNET
Peer-to-Peer Netw. Appl.

Fig. 3 Procedure for nodes


joining a network

receives a join request frame, it will use the distributed consists of two parts: route discovery and route mainte-
address allocation mechanism to judge whether it has ability nance. In route discovery, source node first broadcasts a
to adopt a child node. If available, it will assign a network route request (RREQ) packet across the network via flood-
address and send a join response frame to the child node. ing. Once neighboring nodes receive the packet, each of
Otherwise, it will discard the join request frame. Comparing them judges whether the destination address of the packet
with Zigbee standard network layer protocol, the proposed is its network address. 1) If two addresses match with each
protocol has the following differences in network join pro- other, the node will add this route to its routing table and
cedure. 1) In Zigbee standard network layer protocol, even establishes the reversed pointer to the source in its rout-
if the required parent node has no capacity to adopt the child ing table entries. Next, it sends a route reply (RREP) to the
node, it is still required to send a feedback message. 2) Once source node along the reversed direction. 2) If two addresses
the node is rejected to join, it will carry out the channel scan are not matched, then the node searches its routing table to
in the next round. However, this action is neither necessary find a possible route to the destination. If the route exists,
nor energy-efficient because it is still possible that a proper the node sends a join response frame to the source node,
node in its neighbor table can be its parent node. In addition, and sends a message to the destination. Otherwise, the node
these actions will increase the traffic load and cause more establishes the reversed pointer to the source in its routing
transmission conflicts. In Section 4, the experimental results table entries and then continues to flood the RREQ packet.
demonstrate that our proposed protocol can save more than Note that device nodes employ the destination address
30% communication overheads for network join procedure. and broadcast serial number of the source node as the
unique identifier to avoid repeatedly broadcasting RREQ
2.4 Route discovery and maintenance packets.
In route maintenance, every router node needs to main-
AODV [19] routing algorithm is an on-demand algorithm, tain its own routes to guarantee their validness after they are
which builds up routes between source nodes and destina- established. Thus, every node should periodically broadcast
tion nodes only when it desires to. It uses sequence numbers a hello message to determine whether the current routes are
to avoid occurring routing loops. valid. If a route becomes invalid, it will broadcast a route
In this paper, we propose an improved AODV rout- error (RERR) message to inform the source that the route is
ing algorithm. Similar to AODV, the proposed algorithm now unreachable to destination(s).
Peer-to-Peer Netw. Appl.

The routing process of the improved AODV routing algo-


rithm in our proposed model is described as follows. If an
end device intends to send a data packet, it sends the data
packet directly to its parent node. Note that the parent node
must be a router or the coordinator. For a router or a coordi-
nator, it sends the data packet directly to the next hop node if
it has a route to the destination. Otherwise, it initiates a route
discovery. The overall flow chart of the proposed routing
algorithm is given in Fig. 4.
Fig. 5 A diagram of mobility support for end devices

3 Mobility support in the Zigbee-based WSNs

3.1 Mobility support for Zigbee router adaptive routing strategy is proposed for end devices. For
implementing the function of neighbor discovery, a router
In mobile Zigbee-based WSNs, routers (or the coordina- is required to periodically broadcast hello packets. Accord-
tor ) do not need to dynamically change network addresses, ing to these received packets, end devices can generate and
because they have the capability to maintain and repair their maintain their own neighbor tables, and obtain the infor-
route tables. By adopting the improved AODV algorithm, mation of routers within its neighborhood. Obviously, all
routers may reduce the negative effects of node mobil- routers can act as its potential parent node. When the end
ity. When a router fails in the network, it is not required device leaves its parent node, it selects one from these
to change the network address of the router after being candidates as its parent node to forward its data packets.
assigned an initial network address. Figure 5 illustrates the procedure to support the node
mobility for end devices. In Fig. 5, node D is a child of node
3.2 Mobility support for end devices A in area 1 at the beginning. If node D moves to area 2, it
will become out of the communication range of the node A.
For mobile WSNs, when end devices move outside their In area 2, node D can receive HELLO messages from node
parent nodes’ communication range, according to Zig- C to update its neighbor table. Therefore, it can select node
bee/IEEE802.15.4 Standard, they are required to find new C as its parent node. Following this procedure, end device
parent nodes and change their current network addresses. does not need to search a new parent node in order to get a
In order to solve this drawback of the current Standard, an new network address.

Fig. 4 Flow chart of the


improved AODV algorithm
Peer-to-Peer Netw. Appl.

Compared to the traditional AODV protocol, our pro- Table 2 The number of essential routers
posed routing protocol is superior in following two aspects.
Network size ZMOMSL ZCNSMOS
1), the traditional AODV protocol does not distinguish dif-
ferent types of devices, so as to result in broadcasting 20 6.15 2.75
redundant RREQ packets for end devices during route dis- 40 13.35 5.65
covery phase and aggravating network congestion. 2), the 60 19.50 10.45
conventional AODV protocol can not provide any support 80 25.90 16.35
for the mobility of three types of nodes. 100 33.85 24.05

4 Simulation and analysis


ZMOMSL. When n increases to 100, this ratio can be fur-
In this section, we conduct a comparative performance ther reduced to nearly 14 . Since in general router is a full
evaluation between our proposed Zigbee compliant simula- function device with routing capability and is much more
tion model using the OPNET simulator (ZCNSMOS) and expensive than end device, our proposed ZCNSMOS has a
ZMOMSL, in terms of the number of essential routers, clear advantage in terms of networking cost.
networking overhead, and network join delay. In order to In order to evaluate the energy efficiency of two models
assess validity of the proposed model for supporting node in networking, we demonstrate the communication over-
mobility, we conduct experiments to measure the time taken head required for all nodes joining a network. For compar-
by nodes they establish new routes after previous ones ison purpose, we use the number of ACK frames including
fail. All experiments are completed in OPNET Modeler receiving joining request packets and joining response pack-
14.5. In experiments, all device nodes are uniformly scat- ets as a performance metric, since such number can effec-
tered over a 100 × 100 m2 square area, and the number of tively reflect the total communication overhead. Figure 6
device nodes varies from 20 to 100. The commonly used illustrates the relationship between network join overhead
Random Waypoint mobility model is employed and the and network size. As expected, ZCNSMOS has lower net-
maximum velocity is set at 3m/s. Environmental variables working overhead than ZMOMSL for all scenarios under
applied to performance evaluation are shown in Table 1. For consideration, it is because ZCNSMOS is designated to
each experiment, we randomly generate network topologies, make network join procedure more simplified.
repeat 20 runs, and calculate the average value. Figure 7 illustrates the runtime with respect the number
We first test the number of essential routers of two mod- of nodes attached to the network. The network size n is fixed
els for different network size so that all nodes can join at 80. For different values of n, similar conclusions can be
the network. The results are displayed in Table 2. We can obtained. From this figure, we can see that in ZCNSMOS,
observe that with the increase of the network size, denoted all device nodes can quickly join the network once they
by n, the number of essential routers of both ZCNSMOS are deployed. While for ZMOMSL, device node attachment
and ZMOMSL increases, however, the value of ZCNSMOS network happens after nearly 6s. It is because in ZMOMSL,
is always lower than that of ZMOMSL. When n=20, ZCNS- a large number of broadcast packets, which are generated
MOS needs less than half essential routers compared to by all un-joined nodes, lead to traffic congestion and high

Table 1 Environmental variables

Deployment area 100 × 100 m2


Network size 20∼ 100
Transmission range 30m
Transmission power 0.05W
Packet size 512bytes
Packet interval Poisson(10)s
Simulation duration 600s
Mobility model Random waypoint
maximum velocity 3m/s
Packet start time 20s
The maximum number of child nodes 7
The maximum number of routers 5
Fig. 6 The communication overhead of both ZCNSMOS and
The maximum depth 5
ZMOMSL in networking
Peer-to-Peer Netw. Appl.

To evaluate the effectiveness of our proposed model


ZCNSMOS in terms of node mobility support, we mea-
sure the length of intervals between route failure occurrence
and route recovery. In this experiment, the number of end
devices is fixed at 20, and the number of routers, denoted
by λ, varies from 6 to 21. For different amounts of routers,
the average values of time interval lengthes, the high val-
ues and the low values are presented in Fig. 8. It is obvious
that for any values of λ, all time intervals have lengthes less
than 0.1s. Compared to sensing intervals in conventional
applications, which is 10s in our experiments, such time
lengths are rather trivial and can be neglected. These results
clearly demonstrate that ZCNSMOS can quickly establish
new routes for device nodes when their old routes can not
be maintained due to node mobility. With the increase of
λ, new route discovery time decreases. When λ becomes
Fig. 7 Comparison of ZMOMSL and ZCNSMOS in terms of network 16, the average value is close to 0. This is because for end
join delay
devices, statistically, the increase of λ means that its neigh-
bor table is large so that there are more routers within its
packet loss in the network. Then, according to the MAC pro- neighborhood. For routers, the increase of λ means the aris-
tocol, a random waiting time is required for device nodes to ing of the possibility that they have routes to destinations.
retransmit data packets. For any device node, ZCNSMOS
not only makes it more rapidly to join and leave a net-
work, but also gives mobility support. After the node moves 5 Conclusion
to a new area and is required to rejoin a network, shorter
network join delay can help device nodes participate in a In this paper, an OPNET-based simulation model, called
network task more timely. As shown in above experimental Zigbee compliant new simulation model using the OPNET
results, it is apparent that ZCNSMOS enhances the perfor- simulator (ZCNSMOS), is proposed to achieve node mobil-
mance significantly compared to ZMOMSL for the function ity support for Zigbee sensor networks. As the MAC layer
of network join of the network layer. of Zigbee networks has been defined by IEEE 802.15.4

Fig. 8 Time interval of new 1 1


high high
route discovery in different size average average
of sensor networks 0.8 low 0.8 low
time interval (s)
time interval (s)

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
runtime (s) runtime (s)

(a) (b)
1 1
high high
average average
0.8 low 0.8 low
time interval (s)
time interval (s)

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
runtime (s) runtime (s)
(c) (d)
Peer-to-Peer Netw. Appl.

Standard, we focus our work on the framework design 13. Anantdeep E, Sandeep KE, Balpreet KE (2010) Mobile Zigbee
and the implementation of the network layer. The imple- sensor network. J Comput 2(4):95–99
14. Harsh D, Atishay J, Karun V (2010) Impacted of coor-
mentation procedure of the key functions including net-
dinator mobility on the throughput in a Zigbee mesh
work join, routing discovery and maintenance are inves- network. In: Proceedings of 2010 IEEE 2nd international advance
tigated in details. After that, we investigate the network computing conference (IACC2010). IEEE Computer Society,
routing protocols for routers and end devices, as to sup- Piscataway, pp 279–284
15. Dusn S, Natalija V (2008) Performance of IEEE802.15.4 in wire-
port node mobility. Our experimental results show that
less sensor networks with a mobile implementing various mobility
the proposed model can achieve significant performance strategies. In: Proceedings of 33rd IEEE conference on local com-
improvement in networking, routing, and node mobility puter networks (LCN2008). IEEE Computer Society, Piscataway,
support. pp 680–688
16. Marghescu C, Pantazica M, Brodeala A, Svasta P (2011) Simula-
tion of a wireless sensor network using OPNET. In: Proceedings
of 2011 IEEE 17th international symposium for design and tech-
Acknowledgments This work was supported by the National Nat- nology of electronics packages (SIITME 2011). IEEE Computer
ural Science Foundation of China (Grant Nos. 61462021, 61262074), Society, Piscataway, pp 249–252
Opening Project of Guangxi Key Laboratory of Trusted Software 17. Lucio GF, Paredes-Farrera M, Jammeh E et al (2003) OPNET
(Grant No. PF130549), the Natural Science Foundation of Guangxi modeler and NS-2: comparing the accuracy of network simulators
(Grant No. 2012GXNSFAA053224) and the Nature Science and Engi- for packet-level analysis using a network testbed. In: Proceedings
neering Research Council of Canada (NSERC) Discovery Grant. of the 3rd WEAS international conference on simulation. Greece,
pp 700–707
18. Hammoodi IS, Stewart BG, Kocian A et al (2009) A comprehen-
sive performance study of OPNET modeler for Zigbee wireless
sensor network. In: Proceedings of the 3rd conference on next gen-
References eration mobile applications. IEEE Computer Society, Piscataway,
pp 357–362
19. Perkins CE, Royer EM (2002) Ad Hoc demand distance vector
1. Yu F, Jain R (2015) A survey of wireless sensor network sim- (AODV) routing. IETF Internet draft,draft-ietf-manet-aodv-10.txt
ulation tools. http://www.cse.wustl.edu/jain/cse567-11/ftp/sensor/
index.html. Accessed 15 March 2015
2. Riverbed Technology (2015) http://www.riverbed.com/. Accessed
15 March 2015 Xiaolong Li received the B.E.
3. ZigBee Alliance (2005) Zigbee specification V1.0. http://www. degree from the Harbin Insti-
ZigBee.org tute of Technology, Harbin,
4. IEEE Standards Association (2011) IEEE standard for local China in 2003 and the Ph.D.
and metropolitan area networks–Part 15.4: Low-Rate Wireless degree from the Hunan Uni-
Personal Area Networks (LR-WPANs). http://standards.ieee.org/ versity, Changsha in 2008.
about/get/802/802.15.html Since July 2008, he has been
5. Zigbee Alliance (2015) http://www.ZigBee.org. Accessed 15 with the Guilin University
March 2015 of Electronic Technology,
6. Kucuk K, Bandirmali N, Kavak A et al (2013) A modified sectoral Guilin China. In 2013, he won
sweeper-based localization estimation and its implementation in the best paper award at the
a multi-hop wireless sensor networking environment by using ChinaCom conference. He
OPNET. Simulation 89(6):746–761 is currently a Professor. His
7. Shrestha PL, Hempel M, Qian Y et al (2013) Performance research interests include sen-
modeling of a multi-tier multi-hop hybrid sensor network pro- sor networks, M2M networks,
tocol. In: Proceedings of 2013 IEEE wireless communica- and ad hoc networks.
tions and networking conference. IEEE, Piscataway, pp 2345–
2350
8. Ota K, Dong M, Cheng Z et al (2012) ORACLE: Mobility con-
trol in wireless sensor and actor networks. Comput Commun
35(9):1029–1037 Meiping Peng received the
9. Dong M, Kimata T, Sugiura K et al (2014) Quality-of-Experience M.E. degree in gui lin Uni-
(QoE) in emerging mobile social networks. IEICE Trans Inf Syst versity of electronic tech-
97(10):2606–2612 nology, China, Gui lin, in
10. Chen LJ, Sun T, Liang NC (2010) An evaluation study of mobil- 2013. He will pursue a Ph.D.
ity support in Zigbee netowrk. J Sign Proc Syst 59(1):111– at the School of Electronics
122 and Information at the North-
11. Ramyah SR (2012) Load density analysis of mobile Zigbee western Polytechnical Univer-
Coordianator in hexagonal comfiguration. http://www.SciRP.org/ sity, Xi’an, China. His current
journal/wsn. doi:10.4236/wsn.2012.43009 research interests lay in wire-
12. Vlajic N, Stevanovic D, Spanogiannopoulos G (2011) Strategies less sensor network, wireless
for improving performance of IEEE802.15.4/Zigbee WSNs with communication.
path-constrained mobile sink(s). Comput Commun 34(6):743–
757
Peer-to-Peer Netw. Appl.

Jun Cai received the B.S. and Hong Zhang received B.S.
M.S. degrees from Xi’an Jiao- degree in computer science
tong University, Xi’an, China, from JILIN University, China,
in 1996 and 1999, respec- in 2007. And he received
tively, and the Ph.D. degree M.S. degree in information
from the University of Water- technology and telecommuni-
loo, Waterloo, ON, Canada, cations from INHA Univer-
in 2004, all in electrical engi- sity, Korea, in 2010. He is
neering. From June 2004 to currently a Ph.D. candidate
April 2006, he was with Mc- with the Department of Elec-
Master University, Hamilton, trical and Computer Engineer-
On, as a Natural Sciences and ing at University of Manitoba,
Enginnering Research Council Canada. His research interests
of Canada Postdoctoral Fel- include green communication,
low. Since July 206, he has heterogeneous networks and
been with the Department of cognitive radio in wireless
Electrical and Computer Engineering, University of Manitoba, Win- networks.
nipeg, MB, Canda, where he is an Associate Professor. His current
research interests include multimedia communication systems, mobil-
ity and resource management in beyond-third-generation wireless
communication networks, and ad hoc and mesh networks.

Changyan Yi He is cur-
rently a M.Sc. student in
electrical and computer
enginnering, Univeristy of
Manitoba, Winnipeg, MB,
Canada. He received the B.S.
degree from Guilin University
of Electronic Technology,
Guilin, China, in 2012. His
research interests include
radio resource management,
algorithmic game theory, auc-
tion theory and optimization
in wireless Communication
Networks.

También podría gustarte