On the Trade-Off between Computational Complexity and Collaborative GNSS Hybridization

In the last decades, several positioning and navigation algorithms have been developed to enhance vehicular localization capabilities. Thanks to ad-hoc communication networks, the exchange of navigation data and positioning solutions has been exploited to the purpose. This trend has recently suggested the extension of state-of-the art navigation algorithms to the hybridization of independent heterogeneous measurements within collaborative frameworks. In this paper an integration paradigm based on the combination of Global Navigation Satellite System (GNSS) observable measurements is analysed. In this work, a comparison among legacy Extended Kalman Filter (EKF) and a suboptimal Particle Filter (s-PF) is proposed. First we show that under the same assumptions in non-collaborative framework the s-PF easily overcome EKF performances at the cost of a higher computational cost. On the contrary, by analysing a realistic scenario in which a target agent is aided by a set of collaborating peers we showed that a hybridized EKF implementation allows reaching and overcome PF performance at the only expense of network connectivity among few GNSS receivers, while the proposed integration induces minor benefits for an efficient s-PF.