Channel quality estimation and impairment mitigation in 802.11 networks

Wireless communication has been boosted by the adoption of 802.11 as standard de facto for WLAN transmission. Born as a niche technology for providing wireless connectivity in small office/enterprise environments, 802.11 has in fact become a common and cheap access solution to the Internet, thanks to the large availability of wireless gateways (home modems, public hot-spots, community networks, and so on). Nowdays, the trend towards increasingly dense 802.11 wireless deployments is creating a real need for effective approaches for channel allocation/hopping, power control, etc. for interference mitigation while new applications such mesh networks in outdoor contexts and media distribution within the home are creating new quality of service demands that require more sophisticated approaches to radio resource allocation. The new framework of WLAN deployments require a complete understanding of channel quality at PHY and MAC layer. Goal of this thesis is to assess the MAC/PHY channel quality and mitigate the different channel impairments in 802.11 networks, both in dense/controlled indoor scenarios and emerging outdoor contexts. More specifically, chapter 1 deals with the necessary background material and gives insight into the different channel impairments/quality it can be encountered in WLAN networks. Then the thesis pursues a down/top approach: chapter 2, 3 and 4 aim at affording impairments/quality at PHY level, while chapter 5 and 6 analyse channel impairments/quality from a MAC level perspective. An important contribution of this thesis is to undisclose that some PHY layer parameters, such as the transmission power, the antenna selection, and interference mitigation scheme, have a deep impact on network performance. Since the criteria for selecting these parameters is left to the vendor specific implementations, the performance spread of most experimental results about 802.11 WLAN could be affected by vendor proprietary schemes. Particularly, in chapter 2 we find that switching transmit diversity mechanisms implemented in off-the-shelf devices with two antenna connectors can dramatically affect both performance and link quality probing mechanisms in outdoor medium-range WLAN deployments, whenever one antenna deterministically works worse than the other one. A second physical algorithm with side-effects is shown in chapter 3. Particulary the chapter shows that interference mitigation algorithms may play havoc with the link-level testbeds, since they may erroneously lower the sensitivity threshold, and thus not detect the 802.11 transmit sources. Finally, once disabled the interference mitigation algorithm — as well as any switching diversity scheme described in the previous chapter — link-level experimental assessment concludes that, unlike 802.11b, which appears a robust technology in most of the operational conditions, 802.11g may lead to inefficiencies when employed in an outdoor scenario, due to the lower multi-path tolerance of 802.11g. Since multipath is hard to predict, a novel mechanism to improve the link-distance estimation accuracy — based on CPU clock information — is outlined in chapter 4. The proposed methodology can not only be applied in localization context, but also for estimating the multi-path profile. The second part of the thesis moves the perspective to the MAC point of view and its impairments. Particularly, chapter 5 provides the design of a MAC channel quality estimator to distinguish the different types of MAC impairments and gives separate quantitative measures of the severity of each one. Since the estimator takes advantage of the native characteristics of the 802.11 protocol, the approach is suited to implementation on commodity hardware and makes available new measures that can be of direct use for rate adaptation, channel allocation, etc. Then, chapter 6 introduces a previous unknown phenomenon, the Hidden ACK, that may cause frame losses into multiple WLAN networks when a node replies with an ACK frame. Again, a solution is provided without requiring any modification to the 802.11 protocol. Whenever possible, the quantitative analysis has been led through experimental assessments with implementation on commodity hardware. This was the adopted methodology in chapter 2, 3, 4 and 5. Particularly, this has required an accurate investigation of two brands of WLAN cards, particularly the Atheros and Intel cards, and their driver/firmware, respectively MADWiFi and IPW2200, which are currently the most adopted, respectively, by researchers and layman users.