REPUBLIC OF SERBIA

MINISTRY OF DEFENCE

www.mod.gov.rs

MINISTRY OF DEFENCE

Material Resources Sector

Defensive Technologies Department

Military Technical Institute

www.vti.mod.gov.rs

DISTRIBUTED TARGET TRACKING IN CAMERA NETWORKS USING AN ADAPTIVE STRATEGY

NEMANJA ILIĆ

College of Technics and Technology, Kruševac, Serbia and Vlatacom Institute, Belgrade, Serbia, nemili@etf.rs

KHALED OBAID AL ALI

Etimad, Abu Dhabi, UAE and Vlatacom Institute, Belgrade, Serbia, khaled@etimad.ae

MILOŠ S. STANKOVIĆ

Innovation Center, Faculty of Electrical Engineering, Belgrade, Serbia, milsta@kth.se

SRDJAN S. STANKOVIĆ

Faculty of Electrical Engineering, Belgrade and Vlatacom Institute, Belgrade, Serbia, stankovic@etf.rs

Abstract: In this paper problem of distributed target tracking in large scale camera networks using consensus based algorithms is considered. These networks are typified by sparse communication and coverage topologies - the restrictions that motivate modification of the existing widely adopted consensus based distributed estimation frameworks. Several state of the art consensus based algorithms from the literature that take into account these restrictions are inspected, as well as novel adaptive strategy having much weaker requirements for the communication load. Comprehensive comparison of the aforementioned algorithms in view of their state estimation performance based on mean estimation error and disagreement between the nodes’ estimates is given. Differences of the algorithms in terms of communication requirements are discussed in detail.plate.

Keywords: Camera networks, Distributed target tracking, Consensus, Decentralized adaptation.