|
|
Resource Optimization and Quality of Service in Wireless Networks
Adaptive EDCA - Dynamically Adapting EDCA Parameters Based
On Network Conditions
From February 2006 to September 2006, I worked at
Computing Science deptartment,
Umeå University, primarily on resource optimization in WLAN networks, mainly IEEE 802.11/e.
The work was the extension of my earlier work of development of simulation module of WLAN 802.11e EDCA
for the GloMoSim network simulator.
The work included the implementation of the 802.11e EDCA in GloMoSim together
with comprehensive study of service differentiation mechanism of 802.11e EDCA with the help of extensive simulations and wide range of performance metrics. The evaluation phase included some simple simulation scenarios in order to clearly observe the role of
individual EDCA parameters in providing service differentiation, as well as some more realistic real-world scenarios.
The detailed overview can be found here.
The work was continued on the
enhancement of the 802.11e EDCA enabling it to dynamically adapt EDCA parameters based on varying network conditions and
the evaluation of the performance improvements.
As various simulation studies have shown, the default values of the EDCA parameters only yield good performance for limited
number of scenarios. A major weakness of the 802.11e MAC algorithm lies in the fact that the number of collisions dramatically raises
as network becomes congested. There are two major reasons. First, the default CW values (CWmin and CWmax) defined in the standard are
too small for a large number of stations. Second, a certain CW value is only optimal for a particular number of stations, thus as the
number of stations is changed, the CW must be adapted to the new optimal value. Thus given an algorithm which enables the network to
dynamically adapt the parameter values based on network conditions is guaranteed to improve the performance remarkably.
The primary concept of the work is as follows: the AP continuously monitors the network traffic, calculates new EDCA
parameter values using some algorithm, while using certain performance metric as the decision criterion, e.g., throughput, and broadcasts the updated parameters
values in beacon frames. A station, on receiving the beacon frame, updates its parameter values and next time uses new values to contend for the medium.
The functionality of AP has been implemented, the AP maintains algorithms to adapt the appropriate EDCA parameter values according to varying network conditions. A range of algorithms have been implemented to dynamically adapt the parameters. The decision criterion used by the AP to update the parameter values is the accumulated system
throughput. The AP continuously monitors the system throughput. It records throughput at specific intervals, compares the current
system throughput to the previous one and depending on whether the throughput is decreased or increased, it controls the parameter
values. A specific parameter indicates the algorithm the interval (in number of
beacon frames) when the throughput values are compared and
values of EDCA parameters are updated. For example if set to 10, the AP would monitor throughput every
ten beacons and update the parameter values accordingly. Another parameter
Persistence Factor (PF) enables the user to specify with what factor to increase
or decrease the parameter values. For example, PS of 2 indicates that parameter
value is increased to double of the previous value or decreased to one-half.
Various approaches were tested to adapt the EDCA parameter values until the most efficient algorithm was reached on. The final algorithm delivered outstanding results with dramatic improvements in the overall network performance. The
paper based on the work, titled A novel MAC scheme for solving the QoS parameter adjustment problem in IEEE 802.11e EDCA, has been accepted for publication at WoWMoM
2008 (IEEE International Symposium on a World of Wireless Mobile and Multimedia Networks).
| |
|
|
