More (messages) is less (accuracy) in localization

Time-based localization protocols allow a trusted set of nodes called verifiers to determine the location of a node that is not trusted (and possibly malicious), called the prover. Existing literature on secure localization protocols is concerned with defending against various attack models by restructuring the message exchanges and/or cryptographic issues related to the challenge-response dialog. However, in this paper we focus on the effects of timing uncertainties - actual vs the sender's intended transmission time for a message, or actual vs the observer's measured time for a packet arrival or departure event - on the performance of such protocols. In particular, to the best of our knowledge, this is the first paper to study the effect of differences in the underlying message-exchange structure among various Time-of-Arrival (ToA) multilateration protocols on the expected measurement uncertainty when those protocols are implemented on real systems. First, we divide existing ToA multilateration protocols into three classes - SISO, MIMO and SIMO - based on their message exchange structure. We then compare the three classes in terms of the number of messages required and their relative measurement uncertainty. SIMO multilateration requires least number of message exchanges for localization. By analytically comparing the uncertainty and error in measurements, we show that SIMO multilateration is also inherently more accurate than MIMO.