Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

2017 IEEE 14th International Conference on Mobile Ad Hoc and Sensor Systems

Enhanced Stability Routing Scheme against Radio Jamming


Based on NDL in Multi-hop Wireless Networks
Li Lou1, Jian-Hua Fan1, Yong-yang Hu1,LingYun Xu2
1
Nanjing Telecommunication Technology Institute, Nanjing, China
2
Nanjing University of Aeronautics and Astronautics, Nanjing,China
E-mail: louie1000@outlook.com,fjh710119@outlook.com, 18952072065@qq.com, xlyun@nuaa.edu.cn

Abstract—Most of the routing protocols for wireless networks advanced physical layer and MAC layer techniques, such as
have focused on finding a feasible route from a source to a frequency hopping, channel adaptation, error correcting codes,
destination. Routing Protocols in Multi-hop Wireless Networks etc. In order to support better anti-jamming requirements for
(MHWNs) face more serious security and reliability challenges. MHWNs, routing protocols in network layer need to have a
Radio jamming attacks are especially harmful to the reliability of
links, as they can effectively disrupt communication between any
more complex model where the network is characterized with
node pairs result in that links are continuously created and multiple metrics. The goal of anti-jamming routing is to find a
broken. Selecting a long lasting and steady-going route in multi- path that satisfies multiple constraints and still meeting
hop wireless network is a critical task. This paper studies the reliability requirements for MHWNs.
impact of jamming on the common protocols of wireless This paper investigates routing restoration solutions via the
networks and presents an improved reliable AODV routing design of neighbor dependency measured by exponential
protocol based on Neighbor Dependency Level(NDL),which smoothing estimation of CSI between node pairs for multi-hop
named as RPBND protocol. The NDL of a node and its links. Efficient routing schemes can relieve the interference
neighbor is divided into three ranks, that is, fully- caused by the jamming nodes. Therefore we present a link
dependent, half-dependent and independent level. Then, quality based Neighbor Dependency evaluation algorithm to
route selection ways for different NDL are presented, and select best intermediate relaying nodes and a reliable routing
different priorities of the neighbor nodes with different protocol based on Neighbor Dependency level in Multi-hop
NDL to be selected as the next hop node are assigned. The Wireless Networks. According to the link quality evaluation by
performance of the stable routing protocol based on Neighbor Packet Dropped Rate (PDR) and RSSI, neighbor nodes can be
Dependency Level (NDL) is demonstrated in simulation, and partitioned into different Neighbor Dependency level. First,
compared with the traditional Min-hop-based AODV routing based on a realistic representative propagation model and
algorithm and the improved Cost-based AODV routing evaluation results, link model and test results, a Neighbor
algorithm. The simulation results show that the RPBND protocol Dependency Level (NDL) metric is indicated, where selecting
has better data transmission stability. a neighboring node with low NDL as the next hop node will
greatly degrade the transmission reliability.
Keywords—wireless communication; routing stability; anti-
jamming; exponential smoothing estimation; MHWNs Based on this, the NDL of a node and its neighbor is
divided into three ranks, that is, fully-dependent, half-
dependent and independent level. Then, route selection ways
I. INTRODUCTION for different NDL are presented, and different priorities of the
Multi-hop Wireless Network (MHWN) is a special kind neighbor nodes with different NDL to be selected as the next
of wireless communication network with the characteristics of hop node are assigned. Finally, the performance of the reliable
frequently changing network topology[1], which can serve Routing Protocol based on Neighbor Dependency (RPBND) is
autonomous collection of nodes that communicate over tested in our extensive simulation, and compared with some
wireless links through multi-hop fashion. It has been common routing algorithms. The results show that the
envisioned for a variety of applications such as MANETs, RPBND has better data transmission reliability.
Mesh Networks , Cognitive Radio Networks and wireless
II. BACKGROUND AND RELATED WORKS
sensor networks etc. [2][3].
In MHWNs, routing protocols are often concerned with Built upon a shared wireless medium, multi-hop wireless
connectivity, and path selection methods are often according to network is particularly vulnerable to jamming attacks. The
hop-count or delay. Due to the broadcasting nature of the ability to recover routing from attacks is a crucial aspect in the
wireless communication, the data transmission over a channel design of a multi-hop wireless network.
is open to malicious radio interference. Such malicious radio Recent studies in wireless networks have shown that the
interference is often called radio jamming attack. In radio fluctuation of transmission capability of wireless links may
jamming attack, the jammer first identifies the channel in use have a significant effect on the stability and reliability of
and emits random signal over that channel, it could decrease routing protocols [3][4][5]. To promote stability and anti-
the probability of successful broadcasting in the wireless jamming performance of MHWNs, many measures have
communication. To promote stability and anti-jamming focused on advanced physical layer and MAC layer techniques.
performance of MHWNs, many measures have focused on An autonomous cooperative communication scheme with

2155-6814/17 $31.00 © 2017 IEEE 647


DOI 10.1109/MASS.2017.115
optimized physical layer is proposed to improve the nodes with different NDL to be selected as the next hop node
robustness of the upper routing protocol [7], but it relies on are assigned. Finally, the performance of the reliable Routing
advanced transceivers. SSA protocol uses the signal strength Protocol based on Neighbor Dependency (RPBND) is tested
to evaluate link quality, but cannot response to link in our extensive simulation, and compared with the Min-hop-
interruptions expeditiously [8]. In LSPMR routing protocol, based AODV routing algorithm, the Cost-based AODV
by calculating the variance of signal strength of each received routing algorithm. The results show that the RPBND has
data packet, the stability of each link can be predicted [9]. better data transmission reliability.
According to the historical information to evaluate the link We examine that the Channel State Information (CSI)
stability, LSPMR routing protocol cannot dynamically reflect refers to channel properties of a link; it describes how a signal
the current topology variability in the complex
propagates from the transmitter to the receiver and represents
electromagnetic propagation environments. QRME routing
protocol relies on GPS or other auxiliary equipment to get the combined effect of jamming, interference and other
node location, speed and other information in order to implicit attributes of radio, such as scattering, fading, power
calculate the life time of wireless link and choose a stable path decay with distance. In this paper, we propose that link
consequently. The disadvantage of this protocol is that each stability can be evaluated by first exponential smoothing
node needs to be equipped with locating beacon [10]. In PPC- estimation of CSI of links between nodes with their one-hop
MAODV routing protocol, a reliability metric of relative neighbors. And a reliable routing protocol based on neighbor
position change of adjacent nodes i and j was proposed [11]. dependency (RPBND) is proposed in order to get significantly
But these protocols are based on node position judgment and better data transmission reliability of end-to-end.
need external equipments for forecasting mobile trajectory, In summary, the contributions of this paper are as follows:
which should be implemented in special application scenarios. 1. Indicating the existence of different rank of neighbor
A jamming-resilient multipath routing algorithm based on dependency;
Availability History Vectors (AHV) to select fault- 2. Presenting a routing selection criteria according to
independent paths is effective in overcoming jamming impact neighbor dependency;
and maximizing end-to-end availability of these selected paths 3. Validating the routing protocol based on neighbor
[12]. Interference avoidance mechanisms are well concerned dependency by experiments in simulation.
and a proactive multi-path routing mechanism based on new
jamming aware metric is proposed [13]. This mechanisms
need extra support in the form of routing Interference III. ROUTING METRIC AND NEIGHBOR DEPENDENCY MODEL
Activity (IA) entries to build robust anti-jamming paths.
In this paper, we present a reliable anti-jamming routing Some researchers found that link quality is not isotropic,
metric named Neighbor Dependency (ND) integrating local the coverage of power in the field is irregular as shown in
node reliability and link quality in order to select best Fig.1, and some revealed nodes that are geographically close
intermediate relaying nodes and a stable routing protocol to each other may have high spatial correlation in PRRs
based on Neighbor Dependency. (packet receive rates) as shown in Fig2. Some studies in [12]
illustrate the irregular PDR level around sender under the
Due to the lack of comprehensive evaluation of the link influence of jammer.
quality, routing algorithms in MHWNs may result in poor
network performance.
Since the behaviors of node relationships have great
impact on the performance of routing algorithms, we proposed
new link models based on empirical data. According to the
historical and recent information to evaluate the link
relationships, we present a reliable anti-jamming routing metric
named Neighbor Dependency Level(NDL) integrating local
node reliability and link quality in order to select best Fig.1 irregular coverage of Fig. 2 irregular PDR around
intermediate relaying nodes and a stable routing protocol based power in the field Sender under effect of a Jammer in a
on Neighbor Dependency. real communication system

In this paper, we present a reliable anti-jamming routing To focus our study on the stability of routing scheme
metric named Neighbor Dependency (ND) integrating local against jamming, we consider some primary communication
node reliability and link quality in order to select best nodes in MHWNs with limited mobility. The performance of
intermediate relaying nodes and a reliable routing protocol. A routing protocols are greatly depended on the definition of
Neighbor Dependency Level (NDL) metric is indicated, where routing metric. Among them, reliability has to be guaranteed,
selecting a neighboring node with low NDL as the next hop if links are unstable, retransmission or reroute will be typically
node will greatly degrade the transmission reliability. Based time-consuming. The selection of previous relaying node will
influence the same process of subsequent nodes, and there are
on neighbor dependency evaluated by channel state
channel competitions between nodes besides jamming effect.
information, the NDL of a node and its neighbor is divided As a result, the dependency relationship between nodes, which
into different levels. Then, route selection ways for different named Neighbor Dependency Level(NDL) in this paper, is
NDL are presented, and different priorities of the neighbor very complex. It has positive or negative effects on the

648
network. The node dependency is interrelated and coherence. calculate the cumulative NDL factors and choose the best path
It is considered that the relationship between neighbors has based on this information.
been established when there is a successful wireless
transmission link connected with each other. The neighbor One-hop scan
dependency is a description of link connectivity between
neighboring nodes. We divide the node relationships into three The way that each node collects its neighbor information is
levels: (i) fully-dependent, where the link between node and to scan within one hop range periodically. There are two
certain neighbor is often of good quality, stable, and different scan types. Initially, when a mobile device is
symmetric. (ii) half-dependent, where links are of intermediate powered on, it will perform the first scan. The first scan is
quality (in long-term assessment); neighborhood congestion different from future scans in which it must build the neighbor
and neighborhood density are moderate. (iii) independent, list and be ready quickly for participation in real network
where links have poor quality ,neighborhood congestion and activities. This should happen in a relatively short period. Here,
neighborhood density are very high. These links are it is set to two seconds.
inadequate for communication.
After the initial scan, the node enters a normal working
The NDL evaluation algorithm is run in each node to environment where it needs to keep on collecting its neighbor
identify with which neighboring node may construct more information periodically. This time interval should be dynamic,
reliable data transmission link. NDL evaluation algorithm since it depends on how fast the nodes are moving. In an area
consists of two phases, (i) Monitoring Phase and (ii) Steady where most of the nodes are moving at relatively slow speeds
the routing will not change frequently and the scan interval
Phase.
can be set longer to reduce the overhead. However, in an area
In order to periodically check its neighbors’ stability to see
where most of the nodes are running fast, the scan interval
if they are still reliable neighbors, each node broadcasts should be set to a shorter time in order to be more adaptive
periodic beacon messages to its neighbors at every time epoch and quickly respond to network changes. Considering the one
T . NDL does not require beacon messages to include any hop range and the possible moving speed of device, the default
other information except link quality information. Let b j (t ) scan interval in the working condition is set to 10 seconds. If
denote the beacon sent from node j at the t -th interval. Each in the next scan the node compares the new neighbor list with
node i maintains a table denoted by N i , which records the its old one and finds that less than 30% of its neighbors have
changed (this can be caused by either new neighbors coming
active counts of other nodes. The NDL value will be in or old neighbors going away), it will set the scan interval to
propagated along the path and the destination node will 20 seconds; if more than 30% but less than 50% of its
calculate the stability factors and choose the best path based neighbors change, it keeps the time interval as 10 seconds.
on NDL information. However if more than 50% of its neighbors change, it will set
We focus on studying the path selection problem under the the scan interval to 5 seconds in order to gather its neighbor
threat of jamming and give an illustration by the following information more frequently. By this means, each node can
scenario. A mobile jammer will move in the network along the dynamically collect information from its neighbors thus
predefined trajectory and emit signals continuously. It may calculate the neighbor stability factors.
disrupt the normal communication between sender and A simple method for calculating node dependency is
receivers. It has been illustrated that the bit errors per packet expressed as the Node Degree, which is to simply accumulate
and media access delay at receivers are correlated with the the number of one-hop connected nodes observed by the node
movement of the mobile jammer. As a result, it is challenging
to predict the stability of links while mobile jammer is present. i [13] . Degree Centrality ki is proposed and can be
We must give a metric can accurately reflect the dependency n
relationship between nodes and a routing selection policy formulated as ki  k ij , wherein kij  1 if there is an
based on this metric to evaluate individual link reliability and j 1
availability. The access to the dependency relationship available link between node i and j , else kij  0 .Although
between nodes is beneficial to improve the reliability of
routing. Since the characteristics of links have great impact on these methods are convenient for calculation, they cannot
the performance of routing selection, some studies proposed reflect the inner relationships between nodes and mainly
new models to describe the connectivity between nodes based adapt to fixed topology. Therefore, a metric of link stability,
on empirical data. which named as neighbor dependency, is proposed and can be
evaluated by first exponential smoothing estimation of CSI of
NDL routing wireless links between one-hop neighbors. A stable routing
protocol based on neighbor dependency is designed in order to
In monitoring phase, every node periodically scans its one get significantly better data transmission reliability.
hop range and receives these replies from all its neighbors. The Channel State Information will be estimated at the
According to that information, each node sets up a table of its receiver, then quantized and fed back to the transmitter.
neighbors’ stability. While doing routing, the NDL factor will Therefore, the transmitter and receiver may have different
be propagated along the path and the destination node will Channel State Information. In this research, it is assumed that
the bidirectional equivalent of the short distance multi-hop

649
communication networks, they are all represented by CSI. The and  is the weight of statistical mean h stat (i, j , t  1) ,
common form of CSI, Received Signal Strength (RSS), used
to be considered as a poor predictor of link quality[14]. In [15],  and  are all constants between 0 and 1,     1 。
the Channel State Information represented by instantaneous In fact, the channel state information is restricted by the
channel gain matrix elements is obtained as rate of variation of channel quality. In the fast fading system,
  j  N 0 B it is possible to change the channel during the transmission of
hi, j   0    j   ,Pt  0
, (1) only one character, and the instantaneous value of CSI is more
Pt reasonable to describe the channel state. In slow fading
systems, generally referred to multipath induced fading, or due
wherein   j  denotes the instantaneous receiving SINR, to shadowing from obstacles affecting the wave propagation,
N 0 / 2 is the unilateral power spectral density of AWGN the channel sate may be maintained for a period of time after
(Additive White Gaussian Noise), B denotes the estimation of obtaining the real-time CSI, which can improve the
available signal bandwidth and Pt denotes the transmitter communication efficiency by using the statistical mean of CSI.
The actual value of Channel State Information (CSI) can
power. The transmitter power and data rate have been preset be calculated by channel gain which refers to channel
when source transmits a set of short training sequences, so the properties of a link and describes how a signal propagates
distribution of hi, j  is determined by the distribution of the from the transmitter to the receiver and represents the
instantaneous receiving SINR. combined effect of scattering, fading and power decay with
In this paper, we use this kind of Channel State distance, and makes it possible to adapt transmissions to
Information (CSI) metric to represent the current link quality. current channel conditions, which is crucial for achieving
In order to measure it exactly, each node broadcasts a probe stable communication links. The total channel gain is
packet continuously during every epoch with a fixed duration, determined by statistical mean in test window period which
such as every 500 milliseconds. The Window Mean with saved during the routing maintenance process, and next part is
Exponentially Weighted Moving Average (WMEWMA) [16] the instantaneous value of the current moment t .We can adjust
is a receiver-side LQE based on passive monitoring. It the weights of CSI and evaluate the weights according to the
smoothes PRR estimates using the EWMA filter, which fading feature of channel.
provides more stable but sufficiently agile estimation Although the common form of CSI is RSSI [13], used to
compared to PRR. be considered a poor predictor of link quality, this kind of
metric is easy for acquisition, mostly because of few
The probe packet header includes the transmitter power limitations are needed. In this paper, we use this common
value of Pt . According to the received probe packet, each form Of CSI, RSSI, to estimate the weight  firstly, and then
neighboring node will response the detection process in a non we can calculate H i, j  , the EWMA (Exponential Weighted
preemptive mode. And the response message includes the field
Moving Average) estimation of hi, j  .
of instantaneous receiving SINR which can be calculated
according to the aforementioned Equation 1. We depict H i, j  According to sampling sequence X  of RSSI every 500
as the EWMA (Exponential Weighted Moving Average) milliseconds, we use A.B( ) to express the MSE (Mean
estimation of hi, j  in this paper. The value of H i, j  will be Squared Error) of ESV (Exponential Smoothing Value).
quantized at every receiver and fed back to transmitter, and  ( ) is obtained as
then will be saved by transmitter. H i, j  can be used to
evaluate actual value of Channel State Information between  ( ) 
 X   S (1)

(3)
node i and j , H i, j  is used to be considered as a good 
wherein S  X  -1  1   S 1 , S is the
(1) 1 1
predictor of the quality of the link i  j . We divide time
forecast value of RSSI at the th epoch, S 1 is the recent
(1)
into short epochs with a fixed duration Tepoch and define
H i, j  as smoothing value of RSSI at the   1th epoch, and the
H (i, j )  hins (i, j , t )   h stat (i, j, t  1) (2) superscript represents first exponential smoothing.  is the
coefficient of exponential smoothing. By comparing the error
Where h stat (i, j , t  1) is the statistical mean in test window 1
of the predicted value S and the measured value X  under
period (t  1)  Tepoch , hins (i, j , t ) expresses instantaneous different  , the optimal value  with minimal Mean
value of channel state at the t th epoch, it is the basis for the Squared Error can be obtained as
transmitter to adjust the transmit signal in time. It includes the minimize  ( )
distribution of environmental fading and the average channel 
(4)
gain etc.  is the weight of instantaneous value hins (i, j , t ) , subject to 0    1

650
IV. ROUTING MODEL relaying node s with its certain neighboring node x is defined
It is assumed that the existing condition of a stable link as follows.
i  j is that H(i, j)   during the test window period T w , TABLE 1. DEPENDENCY STATUS

where  is the channel gain threshold. Under the same Attribute Declaration
Neighbor_ID( x ) neighbor sequence number , incremental
available bandwidth and SINR, the larger the channel gain is, sequence number, identify the update
the smaller the signal power required for the receiver, which sequence of dependency with certain neighbor
means the link cost is getting smaller. So the weight of x
link i  j , wij can be expressed by H (i, j ) . In the research Neighbor_Address( x) address of neighbor x
, L _Total( x ) times of successful connection with certain
of the minimum cost routing, the weight of link is the neighbor x during test period
forwarding cost of relaying node, and the transmission process L_Break( x) times of disconnection with certain
changes constantly. Considering the aforementioned neighbor x during test period
L_Status( x ) status identification, used to indicate good
conditions, wij is subject to the following separated constraints link or poor one existing between node
s and its certain neighbor x
1/H i, j  H i, j   
wij   (5)
 , otherwise B. Function design of simulation model
Because of random movement of the nodes and jamming By the re-development of the communication node model
effect, the node relationship matrix and the link weight matrix and process model in OPNET simulation platform, some
are both time-variant. Let E denotes the set of links, let S information (such as SINR) can be passed from the physical
and Mac layer to the upper routing layer.
denotes the source and D be the destination. The minimum
Through the improvement design of the function of
cost routing algorithm can be defined as
wlan_physical_layer_data_arrival(), the available channel
min 
( i , j ) E
w ij x ij
bandwidth, signal power, and SINR can be obtained, and the
last one is calculated by the signal-to-interference-plus-noise
  1 i S
  (6) ratio model of pipeline stage. The estimation of received signal
  ij x   x ji   1 i  D power is given in Algorithm 1.
 ( i , j ) E ( j , i ) E  0 otherwise Algorithm 1.Estimation of Received Power
s .t . 

x ij  0 , i , j   E or x ij  1, i , j   E Input:
Ptx: Transmitted Power
 1/ H i , j , H i , j    Prx: Received Power
 w ij   B: Bandwidth
  , otherwise
Lpath_loss:Path Loss
For the routing selection algorithm of mobile multi- Dpath: Propagation Distance
hop networks, these links with best channel state C: Speed of Light
condition and optimal connectivity should be given a λ: Wavelength
higher priority, according to the node dependency. That is, Bband_max: Maximal Band
we do not only need to estimate the cost of link but also Bband_min: Minimal Band
Gtx: Antenna Gain of Transmitter
need to judge its quality, and these will improve the
Grx: Antenna Gain of Receiver
reliability and stability of the preferred subset of next-hop frx_base_freq: Base Frequency of Receiver
relaying nodes. ftx_base_freq: Base Frequency of Transmitter
Output: Estimation of Received Power
V. STABLE ROUTING PROTOCOL BASED ON NODE 1: IF frx_base_freq > ftx_base_freq THEN
DEPENDENCY 2: Bband_min ←frx_base_freq
3: ELSE
A. Basic Protocol Design 4: Bband_min ←ftx_base_freq
Referring to the most common routing protocol of 5: END IF
6: IF frx_base_freq+ frx_bandwidth> ftx_base_freq+ ftx_bandwidth THEN
MHWNs, AODV protocol, the design process includes routing
7: Bband_max ←ftx_base_freq+ ftx_bandwidth
setup process and maintenance process. Firstly, the sender 8: ELSE
transmits routing exploration packets. We propose a new 9: Bband_max ←frx_base_freq+ frx_bandwidth
routing protocol based on AODV, and according to the 10: END IF
feedback message from receivers, the instantaneous channel 11: IF Dpath>0 THEN
state information for routing optimization will be obtained. If 12:
2
Lpath_loss ←λ /(4π·Dpath)
2

the receiver does not have data transmission demand, the 13: ELSE
maintenance package will be broadcast limited in the hop 14: Lpath_loss←1
range by the value of TTL in order to reduce communication 15: END IF
overhead. The basic data structure of Routing Request 16: Prx←Ptx((Bband_max- Bband_min)/ B) Gtx ·Lpath_loss·Grx
17: RETURN Prx
Message is presented. And the dependency status of current
In the MAC process, the corresponding attributes and
their values need to be registered for the calling of upper

651
protocol. The arguments to this procedure form a variable this paper has the best performance of Data Dropped Rate for
length sequence of name-type-value triplet .This procedure the effect of interference on channel state information was
assigns an attribute of a given registered process a value. The estimated by receiver as the link weight during routing
attribute is created for the process if it does not already exist. If selection. This metric reflects the reliability of the routing
it does exist and the type does not match, its type and value are protocol. So the RPBND protocol can ensure more robust path
changed. The process property values are reset in the process in interference environments.
registry for the calling of upper protocol. The important feature
Wireless Network.Load(bits/sec)
of the OPNET process registration mechanism is that any
defined process can register itself in a module level global Load analysis is based on the number of control packets or
process registry. Some custom additional attributes can be routing packets complete the connection establishment
added and the registration process returns a special handle ---- procedure for data delivery in wireless network. If the receiver
the process register handle which is used to access by other is found by sender then in that case it also replies to sender for
processes. sending data. This graph represents the routing load analysis in
case of AODV, CAODV and RPBND in Fig.4.
VI. PERFORMANCE EVALUATION The greater quantities of routing load will degrade the
network performance and affect the data packets delivery. It
In order to evaluate the performance of our approach,
means the overhead in case of AODV are heavier than other
we compare the RPBND(Reliable Routing Protocol Based on
protocols for lack of link quality evaluation. Here we notice
Node Dependency) protocol presented in this paper with the
that the performance of our proposed protocol RPBND is much
classical AODV protocol and CAODV(improved AODV
better than default AODV. Fig.4 shows the result that the
protocol based on Forwarding Cost) in the same network
RPBND technique gives better performance and fulfill network
environment. The experiments are done in the simulated
requirement for considering the link quality. But it needs more
multi-hop network environment built by OPNET Modeler, 16
control packets to maintain channel state information
wireless nodes are randomly deployed over 5000 x 5000 m2
acquisition during idle time, the routing load level is higher
region. Each node has a transmission range of 1000m and uses
than CAODV but much lower than default AODV.
a DCF MAC protocol. The jamming nodes include the fixed
jammer and mobile one with the random movement
characteristics. The working mode of jammer is multi-
frequency, the interference interval is randomly from 1-2
seconds and the interference duration is randomly from 1-4
milliseconds.
Data Dropped Rate
The data dropped rate reflects the performance of the
routing protocol whether can adapt to the network state change.
The interference may cause the retransmission times to exceed
the maximum threshold and result in data dropped. The data
Fig.4. Routing Load analysis
dropped rate Performance of the three routing protocols under
interference circumstances is show in Fig. 3. End-to-end Average Delay
The end-to-end average delay is an important element of
network performance in multi-hop wireless networks. In the
three protocols, the AODV protocol has the highest end-to-end
delay. Without any considering about the link state information,
it simply takes hop counts as the routing selection criterion
which will increase the retransmission delay. The RPBND
protocol greatly reduced the delay caused by the retransmission,
but increased the network coding delay. Although the
complexity of algorithm was heightened in order to obtain the
highly stability, the end-to-end average delay of RPBND was
not increased extremely, and still keep almost the same as the
Fig.3. Data Dropped Rate performance of the three routing protocols
AODV protocols as shown in Fig.5.

Due to the interference effect, the AODV protocol without


consideration of any channel information had the highest data
dropped rate for its retransmission exceeding the maximum
threshold frequently. The CAODV protocol took into account
of the forwarding cost based on link rate, so the data dropped
rate is lower than AODV. The RPBND protocol presented in

652
References 
[1] Mpitziopoulos, D. Gavalas, C. Konstantopoulos, and G. Pantziou. A
Survey on Jamming Attacks and Countermeasures in WSNs[J]. IEEE
Communications Surveys & Tutorials, vol. 11, no. 4, 2009,pp. 42–56.
[2] B.Q. Kan, J.H. Fan.A Novel Jamming-aware Metric for MHWN
Routing[J]. WSEAS Transactions on Computers. Issue 12, Volume 10,
pp.417-426. December 2011.
[3] Lou Li,Fan Jian-Hua,Xu Cheng. Research on Topologically optimized
anti-jamming technology for tactical MANETs. Acta
Armamentarii,37(9):1662-1669.2016.
[4] Sangita Pal, Srinivas Sethi. Selection of Reliable Channel by CRAODV-
RC Routing[J]. Emerging ICT for Bridging the Future −Volume 2,
Protocol in Cognitive Radio Ad Hoc Network. Advances in Intelligent
Fig.5. End-to-end average delay performance of the three protocols Systems and Computing. 251-259.
[5] Pavel Mach, Zdenek Becvar and Robert Bestak. Acquisition of Channel
The simulation results show that the end-to-end average State Information for Routing Purposes in Relay-based WiMAX
Networks[C]. 2009 Eighth International Conference on Networks.170-
delay of PRBND algorithm has not been greatly improved 175.
compared with the AODV algorithm .This is because in order [6] Lei Lei, Dan Wang, Liang Zhou, Xiaoming Chen, Shengsuo Cai. Link
to get a higher reliability, we have to sacrifice the computing availability estimation based reliable routing for aeronautical ad hoc
networks[J]. Ad Hoc Networks 20 (2014) 53–63.
performance. The PRBND algorithm has a higher [7] Thomas R. Halford, Keith M. Chugg,”The Stability of Multihop
computational complexity than CAODV algorithm, resulting in Transport with Autonomous Cooperation,” The 2011 Military
a higher encoding delay. This is the next step in our algorithm Communications Conference - Track 2 - Network Protocols and
Performance,pp.1023-1028,2011.
needs to be improved. However, the performance of data [8] Mohamed Amine Abid ,Abdelfettah Belghith. “Stability routing with
dropped rate and network load of RPBND protocol are all constrained path length for improved routability in dynamic
better than AODV and CAODV algorithms. MANETs,” Personal and Ubiquitous Computing. Volume 15, Issue 8,
pp.799-810, December 2011.
VII. CONCLUSION [9] Xia Hui,Jia Zhi-ping,Zhang Zhi-Yong and Edwin H-M Sha. “A Link
Stability Prediction-Based multicast routing protocol in mobile ad hoc
In this paper, we present an enhanced stability routing networks,” Chinese Journal of Computers,Vol.36,No.5,pp. 927-
936May 2013.
scheme based on node dependency by analysis and utilization [10] Deng Shuguang,Wang jianxin,Chen Jianer. “A QoS Routing Base on
of the estimation of channel state information in MHWNs, the Steadiest Path in Mobile Ad Hoc Networks,” Computer
which can predict the probability of successful forwarding Engineering,Vol.28,No.9,pp.45-47,September 2002.
[11] Hu Jie,Chen Bing,Ma Xiang-nan,He XIAO-JING,"Novel improvement
more objectively. Simulation results have provided detailed based on stable path for MAODV protocol",Journal of Computer
insights into the differences between the RPBND protocol, the Application,Vol.29,No.11,pp.93-96,Nov.2009.
classical AODV protocol and the improved CAODV protocol. [12] Hossen Mustafa,Xin Zhang,Zhenhua Liu,Wenyuan Xu,Adrian Perrig.
Jamming-Resilient Multipath Routing[J]. IEEE transactions on
RPBND and CAODV protocol have better performance and dependable and secure computing. Vol.9,No.6,November/December
RPBND can fulfill network QoS requirement better. Finally we 2012.
validate the routing protocol based on neighbor dependency by [13] S. H. YOOK,H. Jeong, A. L. Barabási and Y. Tu. Weighted
Evolving Networks[J].Physical Review Letters 86, 5835,pp.5835-
experiments in simulation. As the design of RPBND protocol 5858,June 2001.
has improved the routing maintenance mechanism, it can adapt [14] Daniele Puccinell, Martin Haenggi. DUCHY: Double Cost Field Hybrid
to the tactical communication conditions quickly and reduce Link Estimation for Low-Power Wireless Sensor Networks[C].
HotEmNets 2008,June 2-3, Charlottesville, Virginia, USA, 2008.
the retransmission delay caused by frequently link interrupts of
[15] Andrea Goldsmith.Wireless Communications, 2nd Edition.Cambridge
jamming. The simulation results show that RPBND protocol University Press,pp.85-87,2012
can ensure more stable and efficient performance in highly [16] Yingying Chen,Wenyuan Xu,Wade Trappe,YanYong Zhang.Securing
dynamic electromagnetic propagation environments. Emerging Wireless Systems[M],Springer Science+Business Media,
LLC, 233 Spring Street, New York, USA, 217-219.

653

You might also like