CrowdREM: Harnessing the Power of the Mobile Crowd for Flexible Wireless Network Monitoring

Abstract

High-speed mobile broadband connections have opened exciting new opportunities to collect sensor data from thousands or even millions of distributed mobile devices for the purpose of crowdsourced decision making. In this paper, we propose CrowdREM (crowdsourced radio environment mapping), a framework with the specific aim of monitoring and modelling wireless cellular networks. CrowdREM enables operator-independent and highly efficient collection of network performance data along all layers of the communications protocol stack. Such extensive information on network load, spectrum usage, or local coverage can help operators to optimize their networks and service quality and enable improved consumer decision making. In this paper, we introduce the \mbox{CrowdREM} mobile architecture and show first results from a prototype implementation on open-source mobile phones. We demonstrate the versatility of using commodity devices for network and spectrum monitoring, and present the challenges originating from the use of uncalibrated and low-precision measurement equipment. We have acquired an extensive data set from using our prototype implementation in a 21-day measurement campaign covering more than 1,000 hours of measurement data. From this we present and discuss the potential derivation of tangible and relevant network performance and signal quality indicators, which could, e.g., be conducted by independent parties.