Documente Academic
Documente Profesional
Documente Cultură
4
Institute of Information Technology of ANAS,
Azerbaijan Republic
secretary@iit.ab.az
greatly improve the quality of real-time traffic by the approaches that propose to provide service differentiation
elimination of intermediate link breaks. Figure 2 illustrates based on 802.11, by either assigning different minimum
an example network where real-time and best-effort traffic contention window sizes ( CWmin ) , Arbitrary Inter Frame
utilize different routes [16]. Spacing (AIFS), or back-off ratios, to different types of
traffic. These approaches can all provide differentiation;
however, the parameters are typically statically assigned and
cannot adapt to the dynamic traffic environment. This
reduces the usage efficiency of the network [16].
We propose an adaptive scheme to address trade-off. The
basic idea is that, because the state of ad hoc networks can
vary greatly due to mobility and channel interference, it is
advantageous to adjust the back-off behavior according to
the current channel condition.
To achieve service differentiation, as well as to adapt to
the current network usage, we combine the collision rate
Figure 1. Functionality of the framework at IP and MAC layers. and current QoS of flow with the exponential back-off
mechanism in IEEE802.11. To do it, classifies flows into
three types: delay-sensitive flows, bandwidth sensitive flows
and best effort flows. The delay-sensitive flows, such as
conversational audio/video conferencing, require that
packets arrive at the destination within a certain delay
bound. The bandwidth-sensitive flows, such as on-demand
multimedia retrieval, require a certain throughput. The best
effort flows, such as file transfer, can adapt to changes in
bandwidth and delay. Due to the different requirements of
flows, each type of flows has its own contention window
Figure 2. An example of the routes for different traffic. adaptation rule [15].
1) Delay-Sensitive Flows: For a delay-sensitive flow, the
dominant QoS requirement is end-to-end packet delay. To
2.2 Call setup for real-time traffic control delay, the end-to-end delay requirement d must be
When a real-time flow is requested, a call setup process is broken down into per-hop delay requirements. Each hop
needed to acquire a valid transmission path with satisfied locally limits packet delay below its per-hop requirement to
QoS requirement. Call setup also enables effective maintain the aggregated end-to-end delay below d. For this
admission control when the network utilization is saturated. paper, each node is assigned with the same per-hop delay
This requires accurate estimation of channel utilization and requirement, d/ m, where m is the hop count of the flow.
prediction of flow quality, i.e., throughput or transmission d − D (n)
delay. ( n+1) ( n)
CW = CW * (1 + a m ) (1)
The proposed QoS approach is based on model-based d
resource estimation mechanism, called MBRP [19]. By m
modeling the node backoff behavior of the MAC protocol th
Where the superscript n represents the n update
and analyzing the channel utilization, MBRP provides both iteration, D denotes the actual peak packet delay at the node
per-flow and aggregated system wide throughput and during a update period and α is a small positive constant
delay[16]. (α=0.1) [15].
Call setup
Call setup process based on the modified AODV routing 2) Bandwidth-Sensitive Flows: For a bandwidth sensitive
protocol, which can be divided into a Request and a Reply flow, the dominant QoS requirement is throughput, which
phase. In the request phase, the source node sends Route requires that at each node along the flow’s route, the packet
Request messages (RREQ) for the new flow. The RREQ arrival rate of the flow should match the packet departure
packet reaches the destination if a path with the needed rate of the flow.
quality exists. During the reply phase, the destination node
sends a Route Reply message (RREP) along the reverse path ( n +1) ( n) ( n)
to the source node [16]. CW = CW + β (q − Q ) (2)
where q is a threshold value of the queue length that is
2.3 Prioritized medium access
smaller than the maximum capacity of the queue, Q
Communication in ad hoc networks occurs in a distributed represents the actual queue length and β is a positive
fashion. There is no centralized point that can provide constant(β= 1). If Q is larger than q, the algorithm decreases
resource coordination for the network; every node is CW to increase the packet departure rate to decrease queue
responsible for its own traffic and is unaware of other traffic length. If Q is smaller than q, the algorithm increases CW to
[16]. decrease the packet departure rate and free up resources for
In Ad hoc networks, priority scheduling algorithm is other flows. As the queue size varies around the threshold
based on IEEE 802.11[6].Currently, there are several
140 (IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 6, June 2010
( n +1) ( n) ( n)
CW = CW × (1 + γ ( f − F )) (3)
Where f is a congestion threshold for idle channel time,
Figure 3. Markov Chain model for the back-off window size.
F is the actual idle channel time and γ is a positive constant
(γ= 0.1) [15]
In Eq.(5) W denotes the Contention Window size of flow.
When the average idle channel time F is smaller than the
With attention to Eq.(5), probability that the station with
threshold value f, the network is considered congested and
low CW versus the station with big CW obtain the channel
the contention window size of the best effort traffic is
and transmits a packet, is high. Using the above three
increased to avoid decreasing the service level of real-time
contention window adaptation algorithms and Eq.(4),
traffic. On the other hand, if the network is lightly loaded so
ensures that flows dynamically adjust their contention
that the idle channel time is larger than f, the contention
parameters to meet their own QoS needs with attention to
window size of best effort traffic is decreased so that the idle
collision rate.
bandwidth can be utilized [15].
Additional to combine the collision rate with the
exponential back-off mechanisms, we use the follow 4. CONCLUSION
algorithm [16]: Using the above three contention window adaptation
algorithms (1,2,3) and Eq.(4), ensures that real-time flows
r dynamically adjust their contention parameters to meet their
Back−off =Rand [0,( 2 +Rcol * pri )*CWmin ]*Slot _Time (4)
own QoS needs with attention to collision rate. A real-time
where Rcol denotes the collision rate between a station’s flow that did not get its required QoS in the past due to
two successful frame transmissions, and pri is a variable competition from other flows decreases its contention
associated with the priority level of the traffic. window size so that statistically it will have a higher chance
By applying Eq. (4), traffic with different priority levels will to obtain the channel in the future (Eq.(5)). A best effort
have different back-off behavior when collisions occur. Also flow, on the other hand, increases its contention window
traffic with same priority levels will have different back-off size when the network is considered busy and hence releases
behavior when collisions occur, with attention to flow the channel to the real-time flows.
current status. Specifically, after a collision occurs, low The novelty of this model is that it is robust to mobility
priority traffic will back-off for longer, and subsequently and variances in channel capacity and imposes no control
high priority traffic will have a better chance of accessing message overhead on the network and in calculates CW
the channel. size, attention to collision rate and flow’s current QoS.
3. MODEL VALIDATION
In this section, we study the behavior of a station with a References
Markov model, and we obtain the stationary [1] S. Lee, G.-S. Ahn, X. Zhang, and A. T. Campbell.
probabilityτ that the station transmits a packet in a generic “INSIGNIA: An IP-Based Quality of Service
(i.e., randomly chosen) slot time. This probability does not Framework for Mobile Ad Hoc Networks”. Journal of
depend on the access mechanism (i.e., Basic or RTS/ CTS) Parallel and Distributed Computing, Special issue on
employed. Consider a fixed number m of contending Wireless and Mobile Computing and Communications,
stations. In saturation conditions, each station has 60:374–406, 2000.
immediately a packet available for transmission. With [2] G.-S. Ahn, A. Campbell, A. Veres, and L.-H. Sun.
attention to Markov model (Figure 3), the probability τ is: “Supporting Service Differentiation for Real-Time and
[18] Best-Effort Traffic in Stateless Wireless Ad Hoc
Networks (SWAN)”. IEEE Transactions on Mobile
2
τ ( p) = Computing, vol 1, pp.192–207, July- Septemeber 2002.
1 + W + pW ∑ m − 1( 2 p ) i
(5) [3] S. Chen and K. Nahrstedt. “Distributed Quality-of-
i=0 Service Routing in Ad-Hoc Networks”. IEEE Journal of
Selected Areas in Communications, vol 17, pp 1454-
1465, August 1999.
[4] T. Chen, M. Gerla, and J. Tsai. “QoS Routing
Performance in a Multi-hop, Wireless Network”. In
(IJCNS) International Journal of Computer and Network Security, 141
Vol. 2, No. 6, June 2010
Proceedings of the IEEE ICUPC’97,vol 2, pp 557—61 , EEE/ACM Transactions on Networking, vol 1, pp:397–
San Diego, CA, October 1997. 413, 1993.
[5] P. Sinha, R. Sivakumar, and V. Bharghavan. CEDAR: [18] G.Bianchi, "Performance Analysis of IEEE 802.11
“A Core-Extraction Distributed Ad Hoc Routing Distributed Coordination Function", In IEEE Journal
Algorithm”. In Proceedings of the IEEE Conference on on Selected Areas in Communication, Vol 18, pp 535-
Computer Communications (INFOCOM), pages 202– 547, March 2000.
209, New York, NY, 1999. [19] Y. Sun, X. Gao, E. M. Belding-Royer, and J. Kempf.
[6] IEEE Computer Society.”IEEE Standard for Wireless Modelbased Resource Prediction for Multi-hop Wireless
LANMedium Access Control and Physical Layer Networks. In Proceedings of the 1st IEEE International
Specification”. Novermber 1999. Conference on Mobile Ad-hoc and Sensor Systems
[7] I. Ada and C. Castelluccia. “Differentiation Mechanisms (MASS), Ft. Lauderdale, FL, October 2004.
for IEEE 802.11”. In Proceedings of the IEEE
Conference on Computer Communications
(INFOCOM), Anchorage, Alaska, April 2001.
[8] F. Cal´i, M. Conti, and E. Gregori. “Tuning of the IEEE
802.11 Protocol to Achieve a Theoretical Throughput
Limit”. IEEE/ACM Transactions on Networking, vol 8,
pp 785-799, December 2000.
[9] T. S. Ho and K. C. Chen. “Performance Evaluation and
Enhancement of CSMA/CA MAC Protocol for 802.11
Wireless LANs”. In Proceedings of the IEEE
PIMRC,vol 18, pp 535-547, Taipei, Taiwan, October
1996.
[10] H. Kim and J. C. Hou. “Improving Protocol Capacity
with Model-based Frame Scheduling in IEEE 802.11-
operated WLANs”. In Proceedings of the Ninth Annual
International Conference on Mobile Computing and
Networking (Mobi COM’03), pages 190–204, San
Diego, CA, September 2003.
[11] A. Banchs, X. Perez-Costa, and D. Qiao. “Providing
Throughput Guarantees in IEEE 802.11e Wireless
LANs”. In Proceedings of the 18 th International
Teletraffic Congress (ITC-18), Berlin, Germany,
September 2003.
[12] V. Kanodia, C. Li, A. Sabharwal, B. Sadeghi, and E.
Knightly. “Distributed Multi-Hop Scheduling and
Medium Access with Delay and Throughput
Constraints”. In Proceedings of the Seventh Annual
International Conference on Mobile Computing and
Networking (MobiCOM’01), Rome, Italy, July 2001.
[13] R. Rozovsky and P. Kumar. ”SEEDEX: A MAC
Protocol for Ad Hoc Networks”. In Proceedings of the
2 nd __ACM International Symposium on Mobile Ad
Hoc Networking and Computing (MobiHoc’01), pp 67-
75, Long Beach, CA, October 2001.
[14] A. Veres, A. T. Campbell, M. Barry, and L.-H. Sun.
“Supporting Service Differentiation in Wireless Packet
Networks Using Distributed Control”. IEEE Journal of
Selected Areas in Communications, vol 19, pp 2081,
October 2001.
[15] Yaling Yang and Robin Kravets. "Distributed QoS
Guarantees for Realtime Traffic in Ad Hoc
Networks".Technical Report UIUCDCSR-2004-2446,
June 2004.
[16] Yuan Sun, Elizabeth M. Belding-Royer, Xia Gao and
James Kempf. "Real-time Traffic Support in Large-
Scale Mobile Ad hoc Networks" Proc. of BroadWIM
2004, San Jose, CA, October 2004.
[17] Sally Floyd and Van Jacobson. “Random early
detection gateways for congestion avoidance”.