Overhead and Sensitivity to UWB Ranging Models within a Distributed Bayesian Positioning Solution

For the last past years, many algorithmic solutions have been proposed to solve the positioning problem within IR-UWB wireless networks. Among them, distributed Bayesian schemes are likely to offer flexibility and immunity against the harmful effects of NLOS ranging biases on location precision and robustness. However, as these solutions theoretically require the use of refined ranging models, the availability, the representativeness, and the reliability of prior statistical information in real systems still have to be carefully discussed. In this paper, we firstly investigate the sensitivity of distributed Bayesian schemes to assumed ranging models, through the particular example of the cooperative maximization of the log-likelihood of pair-wise range measurements. More specifically, we evaluate the relevance of approximated solutions, as well as the impact of an erroneous or partial prior knowledge of underlying models and parameters, in comparison with optimal estimators under distinct UWB ranging assumptions. Finally, since distributed schemes are expected to reduce relaying constraints and network latency, an estimation of the overhead and traffic generated by the retained positioning algorithm is also provided. For this purpose, we rely on packet-oriented simulations carried out under realistic network/protocol constraints.