23-07-2012, 01:40 PM
Eliminating the Performance Anomaly of 802.11b
[attachment=28500]
Abstract.
In this paper, we propose a mechanism to eliminate the performance
anomaly of IEEE 802.11b. Performance anomaly happens when
nodes that have different transmission rates are in the same wireless cell.
All the nodes in the cell might experience the same throughput even
though their transmission rates are different because DCF of WLAN
provides equal probability of channel access, but it does not guarantee
the equal utilization of the wireless channel among the nodes. To reduce
such a performance anomaly, we adjust the frame size proportionally
depending on the bit rate. Additionally, our scheme eliminates the performance
anomaly in multi-hop case. Simulation study shows that our
scheme achieves an improvement in the aggregate throughput and the
fairness.
1 Introduction
The performance of IEEE 802.11b Medium Access Control (MAC) is a challenging
issue. Especially, on the fairness of wireless channel utilization, many MAC
layer’s fair scheduling algorithms have been proposed [1, 2, 3, 4]. For the long
term fairness, IEEE 802.11b defines Distributed Coordination Function (DCF),
which gives every node a same probability to access the wireless channel.
In this paper, we address the problems in the Automatic Rate Fallback
(ARF) of IEEE 802.11[5]. The basic CSMA/CA is at the root of the performance
anomaly[1] of 802.11b MAC. If there are nodes which works at different
bit rate in the same cell, the 802.11 cell shows the performance anomaly. The
throughput of all hosts that have higher bit rate are degraded, and all the hosts
in the wireless cell experience the same throughput regardless of the transmission
rate. The main reason is that CSMA/CA mechanism does not provide the
same probability to utilize the channel while it guarantees that all the nodes in
the same wireless cell have the same probability to access the channel. In terms
of the channel utilization, this is quiet unfair because the higher bit rate node
defers transmission longer than that of the lower bit rate node.
To eliminate such a performance anomaly, we adopt the Maximum Transfer
Unit (MTU) adaptation scheme. By adjusting the MTU size as to the transmission
rate, all the nodes can fairly utilize the wireless channel. We show that
our scheme achieves higher throughput than that of normal case. Moreover, it
avoids the performance anomaly in multi-hop situation.
P. Lorenz and P. Dini (Eds.): ICN 2005, LNCS 3421, pp. 1055–1062, 2005.
c Springer-Verlag Berlin Heidelberg 2005
1056 S.-h. Yoo et al.
This paper is organized as follows. We introduce the motivation of this study
in Section 1. In Section 2, the ARF of WLAN and its performance anomaly
are introduced. Section 3 presents our analysis which probes that our scheme
achieves higher throughput than existing method. In Section 4, we differentiate
our work from previous work. Simulation results by the NS-2 simulator are shown
in Section 5. Finally, concluding remarks are given in Section 6.
2 Related Work
2.1 ARF of IEEE 802.11b and Its Variants
It is well known that IEEE 802.11 provides multi-rate capability at the physical
layer. ARF defines several transmission rate of 802.11 for temporal degradation
of wireless channel. When wireless channel is bad, the sender changes the sending
rate to lower level.
ARF takes the advantages of positive ACK in explicit link layer. When a
sender misses two consecutive ACKs, it drops the sending rate by changing
the modulation or channel coding method. In contrast, when timer expires or
consecutive 10 ACKs are received successfully, transmission rate is upgraded to
the next higher data rate.
While the ARF mechanism provides good estimation of link condition between
a fixed pair of nodes, it overlooks the fact that actually not the sender
but the receiver need estimation of the channel condition. A direct disadvantage
of ARF scheme can be clearly seen when there are multiple nodes communicating
with each other in a wireless network. If a node moves to a location with a
bad radio characteristics (communication blockage), other nodes communicating
with this particular node may experience transmission failures and consequently
the transmission rate would be dropped.
Holland [2] proposed the receiver based auto rate scheme (RBAR). In their
scheme, the receiver estimates the wireless channel condition and gives the sender
feedback information. The sender takes advantage of feedback information and
selects the sending rate.
Sandeghi [6] proposed the opportunistic media access scheme (OAR) which
extends RBAR. The main idea of OAR is to exploit good channel conditions
to transmit as many as possible while retaining the long term fairness provided
by 802.11b. OAR achieves fairness of channel utilization by sending a burst of
packets for a single RTS-CTS handshake. The number of packets transmitted by
OAR in any transmission burst should be limited so as to provide a fair channel
access to all nodes. The fair temporal share is determined as the maximum time
the channel can be occupied if OAR transmitted a single packet at the base rate.
The base rate of a channel is the lowest possible rate with which data can be
transmitted. For example, the base rate of 802.11b channel is 2 Mbps. Thus the
number of packets sent in every burst is limited to at most 5 packets when the
selected transmission rate is 11 Mbps. This guarantees that OAR inherits the
same temporal fairness properties of the protocols based on original 802.11.
Eliminating the Performance Anomaly of 802.11b 1057
2.2 Performance Anomaly of 802.11b
Heusse [7] showed the performance anomaly of 802.11b. They analyzed the
anomaly theoretically by deriving simple expressions for the useful throughput,
validated them by means of simulation, and compared with several performance
measurements. In their results, the throughput experienced at each node is same
although the data rate is different. The expression for a throughput is as follows:
Xs = Xf = Sd
(N − 1) · Tf + Ts + Pc(N) × tjam × N
, (1)
Where Xf is throughput at the MAC layer of each of the N-1 fast hosts, Xs
is throughput at the MAC layer of slow host. N is number of nodes, and Tf and
Ts is transmission time for fast nodes and slow node for a packet, respectively.
Pc(N) is probability of collision, Sd is frame size and tjam is the delayed time
experienced by collision.
In the above expression, the throughput is not related with the sending rate
of a node because all the nodes have the same transmission time and the same
frame sizes.
[attachment=28500]
Abstract.
In this paper, we propose a mechanism to eliminate the performance
anomaly of IEEE 802.11b. Performance anomaly happens when
nodes that have different transmission rates are in the same wireless cell.
All the nodes in the cell might experience the same throughput even
though their transmission rates are different because DCF of WLAN
provides equal probability of channel access, but it does not guarantee
the equal utilization of the wireless channel among the nodes. To reduce
such a performance anomaly, we adjust the frame size proportionally
depending on the bit rate. Additionally, our scheme eliminates the performance
anomaly in multi-hop case. Simulation study shows that our
scheme achieves an improvement in the aggregate throughput and the
fairness.
1 Introduction
The performance of IEEE 802.11b Medium Access Control (MAC) is a challenging
issue. Especially, on the fairness of wireless channel utilization, many MAC
layer’s fair scheduling algorithms have been proposed [1, 2, 3, 4]. For the long
term fairness, IEEE 802.11b defines Distributed Coordination Function (DCF),
which gives every node a same probability to access the wireless channel.
In this paper, we address the problems in the Automatic Rate Fallback
(ARF) of IEEE 802.11[5]. The basic CSMA/CA is at the root of the performance
anomaly[1] of 802.11b MAC. If there are nodes which works at different
bit rate in the same cell, the 802.11 cell shows the performance anomaly. The
throughput of all hosts that have higher bit rate are degraded, and all the hosts
in the wireless cell experience the same throughput regardless of the transmission
rate. The main reason is that CSMA/CA mechanism does not provide the
same probability to utilize the channel while it guarantees that all the nodes in
the same wireless cell have the same probability to access the channel. In terms
of the channel utilization, this is quiet unfair because the higher bit rate node
defers transmission longer than that of the lower bit rate node.
To eliminate such a performance anomaly, we adopt the Maximum Transfer
Unit (MTU) adaptation scheme. By adjusting the MTU size as to the transmission
rate, all the nodes can fairly utilize the wireless channel. We show that
our scheme achieves higher throughput than that of normal case. Moreover, it
avoids the performance anomaly in multi-hop situation.
P. Lorenz and P. Dini (Eds.): ICN 2005, LNCS 3421, pp. 1055–1062, 2005.
c Springer-Verlag Berlin Heidelberg 2005
1056 S.-h. Yoo et al.
This paper is organized as follows. We introduce the motivation of this study
in Section 1. In Section 2, the ARF of WLAN and its performance anomaly
are introduced. Section 3 presents our analysis which probes that our scheme
achieves higher throughput than existing method. In Section 4, we differentiate
our work from previous work. Simulation results by the NS-2 simulator are shown
in Section 5. Finally, concluding remarks are given in Section 6.
2 Related Work
2.1 ARF of IEEE 802.11b and Its Variants
It is well known that IEEE 802.11 provides multi-rate capability at the physical
layer. ARF defines several transmission rate of 802.11 for temporal degradation
of wireless channel. When wireless channel is bad, the sender changes the sending
rate to lower level.
ARF takes the advantages of positive ACK in explicit link layer. When a
sender misses two consecutive ACKs, it drops the sending rate by changing
the modulation or channel coding method. In contrast, when timer expires or
consecutive 10 ACKs are received successfully, transmission rate is upgraded to
the next higher data rate.
While the ARF mechanism provides good estimation of link condition between
a fixed pair of nodes, it overlooks the fact that actually not the sender
but the receiver need estimation of the channel condition. A direct disadvantage
of ARF scheme can be clearly seen when there are multiple nodes communicating
with each other in a wireless network. If a node moves to a location with a
bad radio characteristics (communication blockage), other nodes communicating
with this particular node may experience transmission failures and consequently
the transmission rate would be dropped.
Holland [2] proposed the receiver based auto rate scheme (RBAR). In their
scheme, the receiver estimates the wireless channel condition and gives the sender
feedback information. The sender takes advantage of feedback information and
selects the sending rate.
Sandeghi [6] proposed the opportunistic media access scheme (OAR) which
extends RBAR. The main idea of OAR is to exploit good channel conditions
to transmit as many as possible while retaining the long term fairness provided
by 802.11b. OAR achieves fairness of channel utilization by sending a burst of
packets for a single RTS-CTS handshake. The number of packets transmitted by
OAR in any transmission burst should be limited so as to provide a fair channel
access to all nodes. The fair temporal share is determined as the maximum time
the channel can be occupied if OAR transmitted a single packet at the base rate.
The base rate of a channel is the lowest possible rate with which data can be
transmitted. For example, the base rate of 802.11b channel is 2 Mbps. Thus the
number of packets sent in every burst is limited to at most 5 packets when the
selected transmission rate is 11 Mbps. This guarantees that OAR inherits the
same temporal fairness properties of the protocols based on original 802.11.
Eliminating the Performance Anomaly of 802.11b 1057
2.2 Performance Anomaly of 802.11b
Heusse [7] showed the performance anomaly of 802.11b. They analyzed the
anomaly theoretically by deriving simple expressions for the useful throughput,
validated them by means of simulation, and compared with several performance
measurements. In their results, the throughput experienced at each node is same
although the data rate is different. The expression for a throughput is as follows:
Xs = Xf = Sd
(N − 1) · Tf + Ts + Pc(N) × tjam × N
, (1)
Where Xf is throughput at the MAC layer of each of the N-1 fast hosts, Xs
is throughput at the MAC layer of slow host. N is number of nodes, and Tf and
Ts is transmission time for fast nodes and slow node for a packet, respectively.
Pc(N) is probability of collision, Sd is frame size and tjam is the delayed time
experienced by collision.
In the above expression, the throughput is not related with the sending rate
of a node because all the nodes have the same transmission time and the same
frame sizes.