Robustness Enhanced Sensor Assisted Monte Carlo Localization for Wireless Sensor Networks and the Internet of Things

With distributed sensor systems commonly found in Wireless Sensor Networks or the Internet of Things, knowing the location sensor data was acquired from is very important, especially in scenarios with mobile sensors. Range-free Monte Carlo Localization based approaches are very energy efficient and do not require additional hardware beyond a radio, which is found on sensor nodes anyways. However, the use of motion sensor data based dead reckoning greatly improves the accuracy of location estimates and increases robustness against faulty or malicious actors within the network. In this work, we propose Robustness Enhanced Sensor Assisted Monte Carlo Localization (RESA-MCL). We show RESA-MCL’s effectiveness with respect to both general localization accuracy and robustness against malicious attacks or malfunctioning nodes. To evaluate and compare our scheme against existing approaches, we introduce three attack models based on malicious anchor nodes. The performance of RESA-MCL is evaluated under these attack models and our approach outperforms existing schemes in both very low and higher anchor node density environments, achieving a localization error of 0.5 with an anchor density of 0.33. Overall, RESA-MCL outperforms comparable approaches at lower anchor densities with up to 48% lower localization error and demonstrates strongly increased robustness against attacks with minimal computational overhead.