Adaptive Channel Allocation for IEEE 802.11 Wireless LAN Ye Zhao and Kin K. Leung Electrical & Electronic Engineering Department, Imperial College, London, United Kingdom {ye.zhao I kin.Ieung}@imperiaJ.ac.uk Abstract - Today, IEEE 802.11 wireless LANs WLANs) are widely deployed throughout the world. It is known that static channel assignments are not suicient for these networks due to traic variations. However, in order to maintain compatibility with the existing protocol standards, it is not possible to employ dynamic channel allocation methods to handle traic-load luctuations. Rather, adaptive assignment schemes seem to be appropriate. In this paper, we enhance the MinMax algorithm for static channel assignment for 802.11 networks proposed earlier. Based on the enhanced MinMax algorithm, we propose here a centralized adaptive channel allocation scheme for 802.11 networks. Computer simulation using the ns2 software reveals that our proposed adaptive scheme achieves a noticeable improvement of data throughput over ixed assignment algorithms, while only a relatively small number of APs require channel changes as a means to reduce service interruption to users under the existing protocol standards. Keyword: IEEE 802.11, channel assignment, channel utilization, interference. 1. Introduction Today, wireless LANs (WLANs) are widely deployed all over he world to meet the growing demand for wireless data services. The IEEE 802.11 technology is particularly attractive due to its maturity and low cost. The 802.11 standrd has evolved into three high-speed versions of the speciication: 802.11a, b nd g [1-3]. Among them, 802.11b networks were vey populr nd they are replaced by 802.11g recently due to improved data rate. As the 802.11 access points (APs) are deployed everywhere, the coverage areas of APs re expected to overlap with each other. As a result, the radio interference between adjacent co-chnnel APs can cause their throughput performance to degrade. This is particularly so because there are only 3 non-overlapped channels available in the 2.4 GHz ISM band used by the 802.11 networks. Moreover, the limited number of available chanels makes this interference degradation become unavoidable in dense WLAN environments. In order to prevent signiicant performance degradation in the network, limited requency chnnels should be assigned to APs in an appropriate and eicient manner. It is well recognized that static channel assignments do not provide the best performance of wireless networks, because traic load (thus interference) in a network usually varies in time. When comparing with static assignments, dynmic chnnel allocation typically yields better performance in terms of interference, throughput and delay at the expense of added complexity in the control mechanisms. Unfornately, in the context of IEEE 802.11 WLAN, the existing speciications do not include ny protocols or control mechanisms to support dynamic channel assignment. Although such may be considered in the uture versions of the standards, it will be diicult to maintain backward compatibility with existing protocol standards. Consider an 802.11 WLAN in an oice environment. When colleagues are gathering for a meeting, network traic at the conference room increases and maintains high until the meeting ends. Although such trafic variation has a time scale on the order of minutes, static channel assignment clerly cannot provide the optimal use of limited radio requencies. On the other hand, some form of adaptive channel allocations where the chanel changes at a rate matching the traic variation is desirable. he same comment also applies to other WLAN operation scenarios including outdoor applications. Furthermore, in contrast to the protocol requirements for dynamic channel allocations at a time scale of msec, if designed properly, adaptive chanel assignments do not require protocol chnges to the existing standards. Of course, to make the adaptive algorithms "compatible" with the existing standards, it is highly desirable for the algorithms to re-assin an only small number of chnnels based on the curent channel assignment, when appropriate changes of traic load (or interference) are detected. These have been the motivation for our work in this paper. We observe that WLANs in typical operating environments are oten managed by a network management system. So, it is natural to add the capability of adaptive channel allocations to the centralized system. Therefore, we shall investigate a centralized, adaptive chnnel assignment algorithm. As a high-level view, APs periodically measure and estimate their traic load and report the estimates to the management system. In um, the latter instructs certain APs to change to new radio requencies at n appropriate time, according to the adaptive algorithm. It is expected that the rate of channel changes is low, a seconday objective of the adaptive scheme is to reduce the need of chanel changes, while attempting to "optimize" the throughput performance. This way,