Towards a fast and stable filter for RSSI-based handoff algorithms in dense indoor WLANs

Abstract

In dense indoor wireless networks, handoffs occur frequently. The criteria to trigger handoffs are not defined by the IEEE 802.11 standard, thus being specific to each manufacturer’s implementation. Current handoff implementations typically use RSSI (Received Signal Strength Indicator) as a performance metric and commonly cause association instability in dense environments, a well-known problem referred to as the ping-pong effect. In this paper, we present a deep analysis of RSSI traces collected in dense indoor environments using the FIBRE testbed. Based on that, we conclude that the RSSI time series presents deep fast fades that occur frequently in bursts of small sizes, which can cause ping-pongs. Motivated by this behavior, we propose a new and simple filtering mechanism called Maximum which targets to eliminate these valleys in the RSSI time series. In a nutshell, this filter chooses the maximum RSSI value from a sliding window containing the last few RSSI samples of the series. We conduct simulations based on real RSSI traces from static and mobile scenarios to evaluate Maximum with respect to other filtering mechanisms found in the literature. Additionally, we present a simplified model of the behavior of Maximum that allows us to study the probability of unwanted handoffs as a function of the RSSI of the available access points. Our analysis reveals that Maximum is able to offer a better tradeoff between handoff triggering delay and stability in mobile scenarios, while also performing well in static scenarios, effectively avoiding the occurrence of ping-pongs in most cases.