You are on page 1of 8

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 7, JULY 2010, ISSN 2151-9617

HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 62

Energy Consumption and Starvation Reduction in


Multi-Robot Wireless Communication
Priya Shanmugam

Abstract—In multi-robot system (MRS), energy consumption and starvation are the major problem while using IEEE 802.11 protocol. The
starvation problem is determined by Contention window, i.e., a communication frame between source and destination. Previous work on
contention window of MAC IEEE 802.11 has a series of window size by doubling their size when it is waiting for the communication channel,
example 8, 16, 24, to 1024. In the proposed work the contention window is incremented and decremented using natural logarithm to achieve
effective communication. By using trapezoidal fuzzy logic the energy consumption between robots used least energy to nearest node and high
energy to farthest node based on the link quality, distance, noise and received signal strength between the neighboring nodes. Thus gets good
routing.

Index Terms—IEEE 802.11, Starvation, Multi-Robot System, Wireless Communication

——————————  ——————————
Potential applications enabled by this technology include
1 INTRODUCTION multimedia Internet-enabled cell phones, smart homes and

T HE study of multi-robot systems has received appliances, automated highway systems, video
increased attention in recent years. A Multi-Robot teleconferencing and distance learning, autonomous sensor
system is comprised of several systems working networks, robotics etc. However, supporting these
together as a whole. It is a multitude of systems each with applications using wireless techniques poses a significant
structure, sensors, actuators & processors. It has a technical challenge. Wireless communication in robot
capability of interacting with each other to fulfil necessary causes high overheads and conflictions.
goals. Ability to communicate and control the robot in a The wireless communication channel is characterized by
multi-robot system is a challenging task. long bandwidth-delay, dynamic connectivity, and error
The issue of communication in multi-robot system has prone transmission. The robots are often equipped with
been studied extensively since distributed system was low-cost low-power short-range wireless network
taken into robotics research. The communication in multi- interfaces, which only allow direct communication with
robot system is divided into two type’s implicit and explicit their near neighbours. Hence, it is virtually impossible for
communication [4]. The distinction between them is, the each node to know the entire network topology at any
implicit communication occurs as a side effect of other given time. Under these circumstances the only practical
actions or “through the world”, whereas explicit approach to distributed command, control and sensing is to
communication is a specific act designed particularly to employ an ad hoc wireless networking scheme.
convey information to other robots on the team. Several In contrast to a WLAN network that needs some base
researchers have studied the effect of communication in towers, an ad hoc network is an infrastructure-less network
multi-robot systems performance in a variety of tasks and where the nodes themselves are responsible for routing
have said that communication will provide the teams of packets. Mobile nodes communicate with each other using
robot to communicate effectively for the given task. multihop wireless links. There is no stationary
In the networked robotics area, researchers try to realize infrastructure, e.g., no base stations. Each node in the
group behaviours found in small insects or animals, in network also acts as a router, forwarding data packets for
order to control and coordinate a team of robots. Especially, other nodes. Furthermore, because the topology in such a
real-time wireless communication can help dynamic network can be highly dynamic, traditional routing
resource management and self-organization for a team of protocols can no longer be used.
cooperative robots. The multiple robots communicate with Here AODV protocol is used for performance evaluation
each other, sharing the same mission, naturally through of a multi-robot system. Consider a group of mobile robots
wireless communications. In this respect, wireless that are required to autonomously disperse throughout a
communication is an excellent candidate for inter-robot region, perform distributed sensing, monitoring or
information exchange. surveillance, and pass the sensed data to a single collection
Wireless communications is a rapidly growing segment point. The robots are most likely equipped with only low
of the communications industry, with the potential to power wireless transceivers whose range is too short to
provide high-speed high-quality information exchange allow direct communication with the data collection point,
between portable devices located anywhere in the world. but sufficient to allow robots to communicate with close
Priya Shanmugam, Pondicherry University, Pondicherry-8.
© 2010 Journal of Computing
http://sites.google.com/site/journalofcomputing/
63 

neighbours. These circumstances qualify an ad hoc wireless samplings of the research that is currently going on in the
networking scheme as a unique choice. field of multi-robot communication system are provided.
The design of ad hoc networks has focused on the Since the advent of high-performance wireless local area
development of dynamic routing protocols that can network (WLAN) at relatively low cost, its use for wireless
efficiently find routes between two communicating nodes. communication among multi-robot has become a practical
The routing protocol must be able to cope with the mobility proposition. In this paper, several communication
of nodes, which often changes the network topology approaches and protocols in multi-robot communication
drastically and unpredictably. domain are discussed. They have also presented some
Next, the IEEE 802.11 MAC protocol, the de facto communication aspects in multi-robot system, such as cost
standard for wireless LANs, includes a distributed evaluation, wireless network layered model and energy-
coordination function (DCF) mode usable for ad hoc optimized protocols. The challenge in these systems is to
network architectures. The IEEE 802.11 MAC protocol is maintain communication even when connections between
designed to provide efficient and reliable frame exchange robots may change dynamically and unexpectedly. This
over the shared wireless channel. Consequently, this work has extended to achieving fault tolerance in multi-
protocol has a rather limited view of the network robot communication, such as setting up and maintaining
conditions. The information available to it is the status of its distributed communications network and ensuring
direct neighbours. communication reliability.
The present work concentrates on communication issues Ulf Witkowski et al [5] proposed a cell- based network
in ad hoc multi-robot wireless communication. Among the with master nodes in each cell forming the basic structure
issues in robotic communication we have taken starvation of the network. Some nodes formed by specially equipped
and energy consumption problem. I.e., when more than robots act as beacons to uniformly span the network. These
one robot tries to communicate using the same channel robots have a role as reference points when positioning
collision may occur that leads to starvation and energy level other mobile robots or humans and at the same time form
may be saved by using least energy for nearest node and the infrastructure to support communication all over the
more energy for farthest node i.e. above 4 or 5 hops (energy search area. They have proposed the cell based approach
consumption problem). and a service discovery protocol. By using this protocol the
The proposed work gives a model that will accommodate robots will be able to efficiently locate resources and
the performance issues of energy consumption and services available in the network.
starvation problem that occur in MAC layer. The framework RoboNet presented by Antonio Chella et
al [6] adopts a composite networking architecture, in which
2 RELATED WORK a hybrid wireless network, composed by commonly
Over the last years considerable progress has been available Wi-Fi devices, and the more recently developed
achieved in the design of control systems for teams of wireless sensor networks, operates as a whole in order to
mobile robots which perform complex tasks like provide a communication backbone for the robots and to
exploration and map building or search and rescue and in extract useful information from the environment. It allows a
Robot Soccer games etc. Osman Parlaktuna et al [1] team of robots to enhance their perceptive capabilities
proposed a sensor based coverage algorithm to multi-robot through coordination obtained via a hybrid communication
system. By considering the energy capacities of mobile network. The purpose of the framework is the coordination
robots they detected the path but does not work well in of a group of robots moving in a structured indoor
dynamic environment. environment in order to manage automatically guided
Jim Pugh et al [2] proposed the algorithm for the control museum tours. Besides the benefits that are expected to
of robot formation. The formation with the triangular lattice emerge from cooperation, a major difficulty arises from the
structure is created using distributed control rules, using fact that all involved entities must achieve efficient
only local information on each robot. In order to make coordination, while acting independently and
formation they concentrated on scalability with respect to autonomously from each other; moreover, an additional
number of robots. There is no evidence that the algorithm difficulty is represented by the need of providing fault
will not work for larger groups of robots although its tolerance to the whole system. The robustness of the
performances might be gracefully decreasing with large communication protocol implemented in the proposed
swarms. framework has been enforced through a fault-tolerant
Pagello et al [3] considers the properties that a multi- leader election mechanism, which allows for an easy
robot system exhibits to perform an assigned task extension process of the innate system knowledge with new
cooperatively. They address the two issues in their paper, information acquired during run-time.
first dynamic role assignment in a team of robots, and Feng-Li Lian et al [7] proposed a dispersion movement
second one deal with the problem of sharing the sensory algorithm for multi-robot systems with simple
information to cooperatively track moving objects. Computation and easily obtainable information. In
Xiao-Lin Long et al [4] presented some wireless addition, a dispatch control rule is proposed based on the
communication schemes and their applications. Broad dispersion algorithm. The dispersion and dispatch control

© 2010 Journal of Computing


http://sites.google.com/site/journalofcomputing/
64 

rules are easy to implement for a practical multi-robot


system to act like a natural creature system. In the future,
the research will focus on the mechanism of adaptively
adjusting of the dispatch control rule, and the theoretical
analysis of the algorithm performance.
An ad-hoc robot networking scheme is more promising
study in today’s world. In Paper [8] they focused on this
scheme by presenting the background of mobile ad hoc
networks, ad-hoc robot wireless communications, and their
applications. And it discusses the protocol stacks and
explains why ad hoc networking is practicable for mobile
robot wireless communications. They also illustrated some
applications for multi-robot systems communication; they
are Robot soccer games, Explosive ordnance or hazardous
materials disposal, Rescue and recovery operations,
Unmanned Vehicles, Planetary and volcano exploration.
Rui Jiang Gupta et al [9] said about the performance
degradation when TCP is used over 802.11b MAC in a
multi-hop wireless scenario. In the presence of a short hop
length connection, the longer connection does not get much Fig 1. Proposed Architecture
bandwidth, so link failure may occur. Other drawbacks are
packet loss, increased delay in link failure detection. For a
static topology or a low mobility environment, increasing 3.1 Starvation Problem
the retransmission limits results in significant improvement The wireless communication scheme used in IEEE 802.11 is
in performance and for a highly mobile environment; such a widely used standard for MAC and PHY layers of
an increase will result in increased delay in link failure wireless LAN. The basic medium access protocol of the
detection. MAC layer of IEEE 802.11 standard is DCF that allows for
Yasushi Kambayashi et al [10] presented a framework for automatic medium sharing between nodes in a WLAN
controlling mobile robots connected by communication through the use of CSMA/CA protocol and a random back
networks using the Ant Colony Optimization. It works well off time. In DCF, all stations and flows are treated with
in static environment but not in dynamic one and it shows same priority and no differentiation mechanism is used to
a good scalability. The big problem arrive in this is collision guarantee packet delay, throughput and traffics [9]. In this
problem and losing of energy when forming the large protocol, whenever a node has to communicate, it first
cluster. senses the medium to determine if another node is
In the proposed work, by using AODV and IEEE 802.11 communicating. If the channel is busy it will put in a back
the problems like energy consumption, scalability and off timer to wait for a particular time i.e., it will wait until
starvation in robotic communication are going to be solved. its contention window is doubled to its maximum value.
This paper also works in tele-operated system and provides After reaching its maximum range, if it still does not get the
an efficient and reliable communication. channel it will discard its waiting state. Thus waiting for a
The rest of the paper is organized as follows. Section III communication channel creates a collision in the network
briefly discusses the proposed work and techniques that leading to starvation problem.
used to avoid starvation, consuming energy and system The MAC layer has two control packets RTS (Request to
performance in wireless communication in multi-robot Send) and CTS (Clear to Send). The source will send RTS
system. The Section IV describes the simulation results that packet to the destination if it is free it will send the CTS
are done in ns2 and Section V brings forth a conclusion. packet. If CTS respond is sent the communication will start
else the contention window is double the size by 64, 128 etc.
Here the Back off time will be chosen randomly i.e., till it
3 PROPOSED WORK gets the communication channel the random time will be
In multi-robot system we are using 802.11 at Mac level, chosen.
AODV at routing level. The proposed work concentrates on In existing MAC layer if two sources fight for a same
issues that occur in multi-robot system. The system communication channel, the one, which has minimum
proposes a design concept for mobile robots to solve the contention window size, will get the channel first. The
issues like starvation and energy consumption problem that other will be put in waiting state. The source, which is in
occur at MAC level, while using robots in wireless communication state, will maintain the contention window
communication. size as minimum till the communication ends. So after
finishing its communication it will get the first chance again
but the one, which was in the waiting state, will double its

© 2010 Journal of Computing


http://sites.google.com/site/journalofcomputing/
65 

contention window size and never get the channel, so, it size of contention window causes the packet collision in the
will starve for the particular communication channel. At network.
last it will discard its communication with that channel. The Fig 2 shows the connection establishing before
The proposed algorithm presents a solution for starting communication. There is n no. of robots. Here we
starvation problem in single communication channel and tested with 21 robots. R21 is the destination robot. Each
sharing the channel to other nodes by incrementing and node will have its own CW size. In the below fig it shows
decrementing the contention window size. that the nodes R1, R7 & R4 trying to access the same
communication channel from R21 by sending the RTS
packet.
The Fig 3 describes that after receiving the RTS packet
the destination node will check the CW size of each node
and send the CTS packet to node that has minimum CW.
Here it sends the CTS packet to R1. The node with
maximum CW size will choose a back off timer and double
its size. The maximum size for CW is 1024 and minimum
size is 32. When the CW reaches Max size it will discard
from the communication. The R1 will decrease its size by
resetting to original value 32 and it will get the channel
again and again, so the nodes R7 & R4 will not get the
communication channel, here the starvation occurs.
From the proposed algorithm the channel is shared by all
the nodes without any collision and starvation problem by
incrementing and decrementing the CW size using natural
logarithm log (x).
Fig 2. Nodes sending RTS packet for accessing the channel The Fig 3 shows that the communication channel is
shared between those two sources and without any
STarve algorithm (ST) collision by implementing the proposed algorithm.
For, incrementing and decrementing
Procedure Starv_solu 
1. /* Let R1 to Rn be the no. of robots */  Incrementing = CW + log (CW)
2. Set  MaxCW,  MinCW,  CW          //  Maximum  Contention      
window (CW), Minimum CW.  Decrementing = CW – log (CW)
3. Set Stime, Maxt                       // Simulation time, maximum time 
for end of simulation. 
4. While (Stime<=Maxt) 
5. N  no.  of  robots  sends  RTS  packet  to  destination  in  single 
channel asking for communication access. 
6. If the destination is not idle sends the CTS packet. 
7. Only  if  (CW=MinCW)  and  send  the  CTS  packet  to  access  the 
communication channel 
8. MaxCW nodes will put in back off timer. 
9. Back off timer increase its CW size by CW + log (CW). 
10. The  MinCW  node,  which  got  communication,  will  reset  its 
value to CW – log (CW) when finishing its communication. 
11. Now the destination will send CTS packet to waiting nodes that 
have MinCW. 
12. The waiting node, which reaches the MaxCW, will discard from 
waiting for the channel. 
13. Do 
14. End Procedure 
Fig 3. Node 21 sends the CTS packet to access the communication to
The 802.11 uses back off timer to access the channel. The the minCW node
Back off timer is fully depends on Contention window
(CW). The CW is increased when it not gets the channel
access. In the present work we increment and decrement In Fig 4. The communication channel is shared between
the contention window size by using natural logarithm node R1 and R7 and R4 is discarded because it reaches the
rather than doubling its size (2n), because by doubling the max value. The throughput for this is measured and shown

© 2010 Journal of Computing


http://sites.google.com/site/journalofcomputing/
66 

in the simulation result and we get a better result. amount of energy. 


9. End procedure 

In the proposed work we are going to apply Ptx


(transmitting Packet power) dynamically according to
distance and noise condition. Here we have used three
types of propagation models, free space, Two- ray ground
and Shadowing model that used in ns2. These three models
are used to predict the received signal power of each
packet. At the physical layer of each wireless node, there is
a receiving threshold. When a packet is received, if its
signal power is below the receiving threshold, it is marked
as error and dropped by the MAC layer.

As the mobile robots are mainly dependent on battery


power, it is important to minimize their energy
consumption. In this proposed work based on the link
quality, transmission power and distance to the destination
the amount of energy is calculated. If the distance is longer
Fig 4. Communication is shared without any starvation
and noisy high energy will be taken and if it is nearer and
no noisy medium or low energy will be taken.
3.2 Energy Consumption We are calculating distance and channel noise level
In multi-robot system as we use more number of robots condition using received packet power (Pr). In the
we need to concentrate on how to conserve the energy Proposed work we are using trapezoidal function to
during multiple tasks. Osman Parlaktuna et al proposed a transmit packet. They are Very Low (VL), Low (L), Medium
multi-robot Sensor based coverage path planning problem (M), High (H) and Very High (VH).
considering robot energy capacities but it’s not an optimal
result. Here we did the energy consumption at MAC level
of 802.11 in MRS and gave the optimal result. In this work
we will see how we get the efficient routing mechanism by VL L M H VH
consuming the energy while transmitting the signals to
nearer or longer distance nodes. In early system they used
the constant packet transmission power. The IEEE 802.11 Fig 5. Trapezoidal fuzzy function for energy Consumption
uses 0.28183815 powers to transmit a packet regardless of Each node maintains a table that contains the predicted
channel condition and hop distance. values. When a node forwards a packet to its neighbour, it
We have proposed an energy consumption (EC) checks its neighbour’s address with the table and applies
algorithm for each node by using the trapezoidal fuzzy sets. that predicted value as transmission power.
Let us see the procedure proposed for this energy In AODV, a new column is added in neighbour table
consumption problem. nb_power (neighbour node power), which is used to know
the neighbour nodes signal strength. When a node sends a
Energy Consumption Algorithm (EC) Route Reply it will cumulate the required power to transfer
a packet from destination to source. Source will select the
Procedure energy_consum  path with lowest power required path.
1. /* Let there be n no. of robots */ 
2. Set  Pr,  Ptx,  L(i),  Req(i)        //  received  packet  power,  Neighbor address Rcvd Power Required Power
transmitting packet power, link, Required energy  
3. Assign  Ptx  dynamically  according  to  distance  and  noise  Table 1. Table maintained by each node to know about its neighbour
level  
4. Each  node  checks  the  energy  required  to  contact  the  The Fig 6 discusses how each node energy power is
neighbour and noise level of the neighbour.  calculated and how much of energy they are transmitting.
First it calculates the distance from the source to
5. Construct  the  trapezoidal  fuzzy  set  with  levels  for  both 
destination, and from that it chooses the shortest distance.
distance and noise as VL, L, M, H, and VH. 
In the fig 5from node A->E->G->K’s distance is smaller
6. From the sets the required energy for each path is calculated. 
when compared to others. From the example given below
7. Total energy required is calculated by 
let as assume, what energy level we can use for the each
  Total energy = Req (i) + Req (i+1)…+Req (n). 
situation,
8. Choose the lowest power required path and send the required 

© 2010 Journal of Computing


http://sites.google.com/site/journalofcomputing/
67 

Let, So this consumes the energy instead of sending same level


to small and large distance path. Based on this set the
In L1, NL=L, D=L, Req1= L energy is transferred and saved. Each node will
automatically use the appropriate energy based on the
In L2, NL=H, D=L, Req2= M trapezoidal fuzzy set.

In L3, NL=VL, D=L, Req3=L


4 SIMULATION RESULT
Where L1, L2 and L3 are links from nodes A->E, E->G In this section, the simulation environment and the
and G->K, NL is the node level, D is the distance, Req is simulation results are discussed. Simulation is done using
required energy level. the network simulator NS-2.

4.1 Simulation Metrics


The metrics used in evaluating the performance are given
in the following.
4.1.1 Throughput:
The throughput is measured for the starvation problem
with the existing Mac layer. The contention window size is
increased and decreased according to natural logarithm.
The experimental result shows better throughput by
sharing the communication channel and reduce the
starvation.
4.1.2 Energy consumed:
Fig 6. Energy consumption The performance analysis is done based on the
parameters Energy consumed and throughput. Energy
Therefore total energy required is calculated by adding values are calculated using the trace file. The analysis done
the required energy level. are collected as a group and then formatted as a tabular
column containing energy conserved by implementing the
Total energy = Req1 + Req2 + Req3 algorithm and the energy conservation is done. The tabular
column is given for 2, 4, 8, 16 and 32 nodes.
From the total energy each node will use only required
energy from the trapezoidal fuzzy set. 4.2 Simulation Profile
The simulation settings are as follows. The network is
No of nodes Total energy consumed (J) designed with 32 nodes within an area of 1000m x 1000m.
Each node moves randomly and has a transmission range
433.420611(without EC) of 250m. The energy consumption is tested with 32 nodes
2 nodes and compared with the existing system and shows the
374.622725(with EC applied) better result.
The Xgraph is taken and the throughput is measured.
438.825193(without EC) The first two-simulation graph shows the throughput
4 nodes
measured by simply doubling the CW size. So the other
438.786083(with EC applied)
nodes that want to access the communication will not get
398.875434(without EC) the communication till the first communication is finished,
8 nodes because the one that starts the communication will reset its
398.847801(with EC applied) value to minimum size when it finishes the communication.
By incrementing and decrementing the CW we have solved
443.959354(without EC) the starvation problem, which occurred in 802.11. The
16 nodes increment and decrement is done by the log(x), the natural
399.921304(with EC applied) logarithm.
The Fig 7 shows the existing graph that has starvation
449.448617(without EC)
problem i.e., the node with minimum CW has got
32 nodes
399.771504(with EC applied)
communication channel and started transmission and
occupied the channel by resetting to minimum value even
Table 2. Total energy consumed for various number of nodes the communication ends. The next Fig 8 shows the
simulation of the node, which does not get the
communication, waits until its CW reaches its maximum

© 2010 Journal of Computing


http://sites.google.com/site/journalofcomputing/
68 

size and discarded from the communication. Thus the


starvation occurs in this.

Fig 10. Communication is shared to another node, which increment its


size, by lower value.

The Fig 9 and 10 shows the result after implementing the


proposed algorithm. It shows that now the single
Fig 7. Communication between first group communication channel is shared between both the nodes
by incrementing and decrementing its CW size. It shows
the better throughput when compared with the existing
one.

From the Fig 11, it is inferred that the energy


consumption is reduced when implemented the proposed
algorithm energy_consum. Thus from the table 2 it can be
clearly viewed that the energy has been successfully
conserved thereby implementing the trapezoidal fuzzy set.

460

440
Total Energy in Joules

420

400 Without EC
380 With EC
Fig 8. The second group which not get the communication channel and
in waiting state. 360

340

320
2 4 8 16 32
Number of nodes

Fig 11. Energy consumption

5 CONCLUSION
The proposed work discusses the problems of starvation
and energy consumption in multi-robot system through
wireless communication channel. A better throughput is
achieved while the system is in starvation by incrementing
and decrementing the contention window size by natural
logarithm instead of doubling its window size. The energy
consumption is calculated based on link quality, distance,
Fig 9. Communication allocated to the smallest CW size node
noise and received signal strength between the
neighbouring nodes.

REFERENCES
[1] Osman Parlaktuna, Aydin Sipahioglu, Gokhan Kirlik, Ahmet

© 2010 Journal of Computing


http://sites.google.com/site/journalofcomputing/
69 

Yazici, "Multi-robot sensor-based coverage path planning using


capacitated arc routing approach”, IEEE Multi-conference on
Systems and Control, Saint Petersburg, Russia, July 8-10, 2009.
[2] Inaki, Navarro, Jim Pugh, Alcherio Martinoli and
Fernando Matia. “A Distributed Scalable Approach to Formation
Control in Multi-Robot Systems”, Proceedings of the International
Synposium on Distributed Autonous Robotic Systems, May-15,
2008
[3] Pagello, E. D'Angelo, A. Menegatti, E., , “Cooperation Issues and
Distributed Sensing for Multi-Robot Systems”, Proceedings of the
IEEE, Publication Date: July 2006, Volume: 9, Issue: 7.
[4] Xiao-Lin Long, Jing-Ping Jiang, Kui Xiang, “Towards Multi-robot
Communication”, Proceedings of the 2004 IEEE, International
Conference on Robotics and Biomimetics, August 22 - 26, 2004,
[5] Ulf Witkowski, Mohamed, E. Habbal, Stefan Herbrechtsmeier,
and Andry Tanoto, Jacques Pendersand , LyubaAlboul, “Ad-hoc
network communication infrastructure for multirobot systems in
disaster scenarios”.
[6] Antonio Chella, Giuseppe Lo Re, Irene Macaluso, Marco Ortolani
and Daniele Peri, “A Networking Framework for Multi-Robot
Coordination”, Recent Advances in Multi-Robot System, I-Tech
Education and Publishing, 2008.
[7] Feng-Li Lian, You-Ling Jian and Wei-Hao Hsu, “Dispersion and
Dispatch Movement Design for a Multi-Robot Searching Team
Using Communication Density “, Recent Advances in Multi-
Robot System, I-Tech Education and Publishing, 2008
[8] Zhigang Wang, Lichuan LIU and MengChu Zhou, “Protocols and
Applications of Ad-hoc Robot Wireless Communication
Networks: An Overview”, Inernational Journal Of Intelligent
Control and Systems, Vol: 10, No. 4, December: 2005, 296-303,
[9] Rui Jiang Gupta, V. Ravishankar, C.V , “ Interactions Between
TCP and the IEEE 802.11 MAC Protocol” , DARPA Information
Survivability Conference and Exposition, 2003.Proceedings IEEE,
Publication Date: 22-24, April 2003, Volume: 1.
[10] Yasushi Kambayashi, Yasuhiro Tsujimura, Hidemi Yamachi,
Munehiro Takimoto, Hisashi Yamamoto, " Design of a Multi-
Robot System Using Mobile Agents with Ant Colony Clustering",
Proceedings of 42nd Hawaii International Conference on System
Sciences, IEEE 2009.
[11] Azade Khalaj, Nasser Yazdani, Maseud Rahgozar, “Effect of the
contention window size on performance and fairness of the IEEE
802.11 standard”, Wireless Pers Commun (2007), Springer Link.
[12] Mayez A. Al-Mouhamed and Umair F. Siddiqi, “Performance
Evaluation of Auctions WLAN for RoboCup Multi-Robot
Cooperation”, Journal of Latex Class Files, September-2008.

© 2010 Journal of Computing


http://sites.google.com/site/journalofcomputing/

You might also like