POVZETEK
Natančna lokacija ljudi in stvari postaja pomembna informacije v novo nastajajočih storitvah in aplikacijah na več področjih, zlasti v konceptu interneta stvari, pri optimizaciji brezžičnih komunikacijskih sistemov, pri zagotavljanju lokacijskih storitev in v različnih aplikacijah, ki uporabljajo brezžična senzorska omrežja. Medtem ko je na prostem ocena lokacije z uporabo satelitskih navigacijskih sistemov dovolj natančna, pa je lokalizacija v notranjosti stavb in v sistemih sestavljenih iz vozlišč z omejenim virom energije še v fazi intenzivnega raziskovanja. V tem prispevku predstavljamo celovit pregled obstoječih metod lokalizacije, primernih za brezžične naprave z omejenimi viri energije, ki so lahko nameščene tako v notranjosti stavb kot tudi na prostem. Podrobneje analiziramo predvsem metode lokalizacije za brezžične sisteme z omejenim številom sider, tj. naprav, katerih lokacija je znana. V tem primeru morajo agenti oziroma vozlišča, katerih lokacija ni znana, za ocenitev svoje lokacije pridobiti dodatne podatke od sosednjih agentov. Množica algoritmov je znana pod imenom kooperativni algoritmi za določanje lokacije. Problem kooperativne lokalizacije je možno rešiti kot nelinearni problem z uporabo metode minimizacije najmanjših kvadratov ali kot problem sklepanja na grafičnem modelu. V delu predstavimo in analiziramo dva kooperativna algoritma za določaje lokacije, in sicer gradientni algoritem s konveksno relaksacijo, ki spada v prvi pristop in algoritem širjenja zaupanja, ki ga razvrstimo v drugi način določanja lokacije. Algoritma smo preizkusili v simulacijskem okolju in z uporabo podatkov pridobljenih iz tesnega okolja Log a Rec.

ABSTRACT
Cooperative localization in wireless sensor networks
The accurate location of peoples and things becomes an important information in emerging services and applications in several areas, in particular in the concept of internet of things, in the optimization of wireless communication systems, in the provision of location based services and in different applications using wireless sensor networks. While outdoors, the location estimation is solved by using navigation satellite systems, the indoor localization and localization of devices with limited power resources is still in the phase of intensive research. In this contribution we give a comprehensive review of existing localization methods appropriate for power limited wireless device, which could be placed indoors and also outdoors. In particular we are interested in localization methods for wireless system with the limited number of anchors, i.e. devices which location is known. In this case the agents, the nodes with unknown location, estimate their location by retrieving additional information from neighbouring agents. This set of algorithms is known as cooperative localization algorithms. The problem of cooperative localization can be solved as a nonlinear least-squares optimization problem or as an inference problem on a graphical model. We present and analyse two cooperative localization algorithms, one applying the first and one for the second approach, namely, a gradient convex relaxation algorithm and a belief propagation localization algorithm. The algorithms are tested in a simulation environment as well as using data obtained from Log a tec testbed.