When maximising the coverage of a camera network, current approaches rely on a central approach and rarely consider the decentralised or even self-organised potential. In this paper, we study the performance of decentralised and self-organised approaches in comparison to centralised ones in terms of geometric coverage maximisation. We present a decentralised and self-organised algorithm to maximise coverage in a camera network using a Particle Swarm Optimiser (PSO) and compare them to a centralised version of PSO. Additionally, we present a decentralised and self-organised version of ARES, a centralised approximation algorithm for optimal plans combining PSO, Importance Splitting, and an adaptive receding horizons at its core. We first show the benefits of ARES over using PSO as a single, centralised optimisation algorithm when used before deployment time. Second, since cameras are not able to change instantaneously, we investigate gradual adaptation of individual cameras during runtime. Third, we compare achieved geometrical coverage of our decentralised approximation algorithm against the centralised version of ARES. Finally, we study the benefits of a self-organised version of PSO and ARES, allowing the system to improve its coverage over time. This allows the system to deal with quickly unfolding situations.
|Name||2017 IEEE 11th International Conference on Self-Adaptive and Self-Organizing Systems (SASO)|
|Conference||11th International Conference on Self-Adaptive and Self-Organizing Systems|
|Period||18/09/17 → 22/09/17|
© Copyright 2017 IEEE - All rights reserved