Location and Network Quality Issues in Local Area Wireless Networks

Abstract

Wireless networks intrinsically contain pockets of lowered quality within the coverage area, where environmental factors cause packet loss, lowered bandwidth, or intermittent connectivity. Early detection of such regions can be extremely valuable; in order to facilitate preemptive or corrective action by protocols and applications, we propose a software framework that detects and predicts impending spots when a mobile device moves through a wireless network. Based on measurements of signal strength, network latency and packet loss, we postulate that mobile devices (and their landline communication partners) can be forewarned with a high degree of accuracy when approaching trouble spots. In this paper we describe a lightweight software framework that assists in monitoring network quality within a coverage area, based on several parameters in isolation and in combination. The effectiveness of using these metrics was measured, and experimental results indicate that accuracies of the order of 95% with very low false positives can be obtained. We conclude with an exemplary outline of how applications may use this software to detect regions of degraded network quality and take compensatory action, resulting in enhanced effectiveness.