Effect of adjacent-channel interference in IEEE 802.11 WLANs Eduard Garcia Villegas, Elena López-Aguilera, Rafael Vidal, Josep Paradells Wireless Networks Group, Telematics Engineering Dept. Technical University of Catalonia (UPC) Barcelona {eduardg, elopez, rvidal, teljpa}@entel.upc.edu Abstract—Frequency channels are a scarce resource in the ISM bands used by IEEE 802.11 WLANs. Current radio resource management is often limited to a small number of non- overlapping channels, which leaves only three possible channels in the 2.4GHz band used in IEEE 802.11b/g networks. In this paper we study and quantify the effect of adjacent channel interference, which is caused by transmissions in partially overlapping channels. We propose a model that is able to determine under what circumstances the use of adjacent channels is justified. The model can also be used to assist different radio resource management mechanisms (e.g. transmitted power assignments) I. INTRODUCTION The number of WLANs keeps growing without control due to the use of unlicensed ISM (Industrial, Scientific and Medical) bands. Moreover, these technologies can be achieved at low costs and interoperability is guaranteed by standardization. In densely populated areas, we can observe the coexistence of enterprise WLANs, public hot spots, wireless domestic users, etc. sharing the same frequency spectrum which is in fact a scarce resource. Described in [1] as a chaotic network, this scenario is characterized by an unplanned and unmanaged deployment. The mentioned study also states that in most cases, the density of nodes is such that administrators are not able to ensure an innocuous coexistence (many interfering sources and a limited number of non-overlapping frequency channels). It is clear that, with an increasing number of neighboring nodes, the undesired effect of interference becomes more problematic, affecting the network performance. Usually, two types of interference are distinguished: co-channel interference, which is caused by undesired transmissions carried out on the same frequency channel; and adjacent channel interference, produced by transmissions on adjacent or partially overlapped channels. The way the nodes of a WLAN share the medium is similar to an Ethernet segment. A CSMA/CA (carrier sense with collision avoidance) is used as medium access control scheme. Nodes sense the air interface before transmitting a frame, if it is busy, they will wait until it be released. This makes the study of interferences in IEEE 802.11 WLANs quite different from what is done in other radio networks due to the particular influence of interferences produced by cells using the same channel (co-channel interference): in a cell suffering only from co-channel interference, even though there is no traffic on it, the nodes may defer their transmissions if when sensing the medium, they detect other nodes using the channel from an interfering cell; these nodes are called exposed terminals The presence of adjacent channel interference reduces the effective SINR (Signal to Interference and Noise Ratio) and therefore, the number of errors in reception is increased. These effects used to be minimized with a good network design, prior to its deployment [2][3]: the best access point locations, transmitted powers and channel allocations are computed off-line in order to obtain the best performance at the same time that the sufficient capacity and a full coverage is guaranteed. IEEE 802.11 networks operate in the 5GHz (.11a) and 2.4GHz (.11b/g) unlicensed frequency bands. Communications in these bands need to implement spread spectrum techniques and limit their transmitted power in order to minimize the impact of interference with other devices. Once spread, the resulting signal occupies a bandwidth of about 20 MHz. In addition, the available channels are defined with 5MHz separation between consecutive carriers, bringing the need to use, at least, five channels of separation to guarantee that two simultaneous transmissions do not interfere with each other. Consequently, whereas there are up to 19 (12 in USA) non-overlapping channels in the 5GHz band, in the 2.4GHz band only three out of 13 (11 in USA) are non- overlapping (traditionally, channels 1, 6 and 11) as shown in fig. 1. Bearing in mind the scenarios mentioned before, where the density of nodes can be very high, only three channels are not enough to guarantee an innocuous coexistence between different WLANs. Previous empirical studies stated that a separation of four channels can be used without reducing the performance [4], so the possibilities could be opened to channels 1, 5, 9 and 13 (where available). The idea of using all available channels appears in [5] for the first time. In this paper we present an analytical study on the effects of adjacent channel interference in IEEE 802.11abg WLANs which is supported by practical measurements and simulations. The