About the practicality of using partially overlapping ... - CiteSeerX

About the practicality of using partially overlapping channels in IEEE 802.11 b/g networks Michael Doering, Łukasz Budzisz, Daniel Willkomm and Adam Wolisz {doering, budzisz, willkomm, wolisz}@tkn.tu-berlin.de Abstract—IEEE 802.11 WLANs are currently one of the most popular wireless technologies, but their immediate success results in dense deployments and high demand of user traffic. This in turn leads to decrease in throughput and poor spectrum utilization. Especially in the 2.4 GHz ISM band, where the spectrum is a very scarce resource, all available WLAN channels should be exploited in the best possible way to achieve higher utilization. One way to reach this goal is the usage of partially overlapping channels (POC). Most of the previous work related to POC is based on two major studies addressing 802.11 b, but none of them evaluates the POC behavior in the 802.11 g networks. Moreover, most of the previous results are based on simulations. The main contribution of this work is an experimental evaluation of POC in 802.11 g networks. In this paper we confirm quantitatively that 802.11 b reacts as expected from the previous studies, while 802.11 g reacts entirely different to the presence of adjacent channel interference. That leads to the conclusion that the usage of POC for 802.11 g is not recommended. Index Terms—WLAN, IEEE 802.11 b/g, partially overlapping channels, interference, channel allocation

I. I NTRODUCTION In the last decade WLAN has become one of the leading wireless communication technologies. Due to the low cost and high availability the IEEE 802.11 b/g hardware is omnipresent, especially in residential environments. IEEE 802.11 b/g is operating in the 2.4 GHz Industrial, Scientific, Medical (ISM) band. In this band the number of WLAN channels is very limited. The IEEE 802.11 b/g standard defines WLAN channels in an overlapping manner and only three of them are orthogonal, i.e., their spectrum mask do not overlap. Most WLAN channel allocation schemes suggest the usage of these non-overlapping channels to avoid transmission problems in areas with high traffic and dense networks. The interference between adjacent WLAN networks is dependent on geographical distance, link utilization, as well as channel separation. Based on these factors, one way to improve the spectrum utilization is the usage of partially overlapping channels (POC). At first glance the idea seems very promising, as current modulation schemes like Orthogonal Frequency Division Multiplexing (OFDM) and Direct Sequence Spread Spectrum (DSSS) are robust to interference. Today the 802.11 b standard has been mostly replaced by 802.11 g due to the higher throughput of the latter. To the best of our knowledge none of the recent studies of POC, either by simulation or by experiment, has addressed the usage of this concept in 802.11 g networks. In addition, even for

802.11 b the experimental results are limited. Therefore, we experimentally confirmed the previous results for 802.11 b and investigated the application of the POC concept for 802.11 g networks. From the standards perspective, 802.11 b and g differs in the underlying modulation scheme, which is Direct Sequence Spread Spectrum (DSSS) for b and Orthogonal Frequency Division Multiplexing (OFDM) for g, respectively. According to the different modulation schemes also the implementation of the channel access scheme (CSMA/CA) differs for both PHYs. All these factors will be studied in this paper. The rest of this paper is organized as follows. Sec. II describes the related work in the context of POC. Sec. III gives a brief overview of the hardware used for measurement. In Sec. IV the measurement methodology is explained. The results are discussed in Sec. V. In Sec. VI the conclusions are given. II. R ELATED WORK The concept of POC has been originally suggested by Mishra et al. in [1], [2]. The authors perform testbed measurements to evaluate how POC can improve throughput and spatial re-use in 802.11 b; further they present an analytical model for POC based on their measurements. Mishra et al. calculate the autocorrelation among WLAN channels of the 802.11 b transmission mask defined by the standard and call it I-factor (interference factor). The I-factor indicates the normalized Signal-to-Noise Ratio (SNR) of adjacent channels and is used as interference model for their simulations and calculations advocating the benefits of POC. In addition, their simulation results show that exploiting POC can increase more than twofold the end-to-end throughput. More experimental data has been provided by Ding et al. [3]. They replicate the measurement scheme proposed by Mishra et al. in [1] observing similar results and additionally calculate a metric called IF (interference factor). The IF represents the throughput of the interfering links in a qualitative manner and is used to create a weighted coloring interference graph to minimize the interference between adjacent links. A second group of studies, related to channel assignment schemes for POC, is based on channel overlap calculations made by Burton [4]. The difference between Burton’s calculations and the interference factors, discussed above [1], [2], [3], is the assumption of another spectrum mask. Whereas Mishra et al. use the transmit spectrum mask for 802.11 b DSSS given in the standard, Burton takes a measured transmit spectrum

c 2013 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

mask after the channel filter was applied. This latter mask is thus narrower. Finally to improve the spectrum utilization Burton suggested in [4] a frequency reuse plan for channels 1, 4, 8 and 11, which is equal to a channel separation of three. This plan is related to 802.11 b interference calculations from his white paper. Zhou et al. [5] suggest a channel assignment scheme for 802.11 b using a Signal to Interference-plus-Noise Ratio (SINR) interference model. This model is also based on the above mentioned channel overlap calculations for 802.11 b proposed by Burton [4]. Feng and Yang [6], [7] develop an analytical approach for interfering networks, in function of different channel distances. Their approach is based on the experimental setup, introduced by Mishra et al. [1], using two interfering pairs of APclient. The authors in [6] and [7] assume, that the distance between the AP and the client in each pair is much smaller than the distance between the interfering pairs (B