This paper evaluates the accuracy of several RSSI-based localization techniques on a live jobsite and compares them to results obtained in an operating building. RSSI-based localization algorithms were tested due to their relative low cost and potential for accuracy. Four different localization algorithms (MinMax, Maximum Likelihood, Ring Overlapping Circle RSSI and k-Nearest Neighbor) were evaluated at both locations. The results indicate that the tested localization algorithms performed less well on the construction jobsite than they did in the operating building. The simple MinMax algorithm has better performance than other algorithms, with average errors as low as 1.2 m with a beacon density of 0.186/m2. The Ring Overlapping Circle RSSI algorithm was also shown to have good results and avoids implementation difficulties of other algorithms. k-Nearest Neighbor algorithms, previously explored by other construction researchers, have good accuracy in some test cases but may be particularly sensitive to beacon positioning.

Localization; Civil construction/buildings; Radio frequency; RSSI; Model based methods


No posts