Urban mobility sensing fortraffic using sparseprocessing

Abstract: This thesis was aimed at studying the existing methods for origin-destination(OD) estimation problem and developing a new algorithm which provideshigher promise.The performance was evaluated on a simulated data-set for Stockholmcity. Data for this study were obtained with the help of G. Fl ̈otter ̈od fromDepartment of Transport Science in KTH. Information minimizing approachand entropy maximizing approach, which are the state-of-art methods intransport field were modified to implement. Several existing algorithms insignal processing field, such as BP/BPDN, OMP and SP, were implementedand analyzed. A recently proposed algorithm calld OMP + was described.Then a more effective method SP + with better reconstruction performancein sparse signal processing area was proposed in this report.By numerical experiments, it was concluded that the methods in signalprocessing field could deal with OD estimation problem well. Hopefully thisthesis could make a contribution to opening the door to another field andintroducing methods of that universe, as well as developing a new algorithmwith robust results and small computation cost.