A network sensor location procedure accounting for o–d matrix estimate variability

The paper illustrates an innovative and theoretically founded methodology for solving the network sensor location problem (NSLP), explicitly accounting for the variability of the o–d matrix estimate. The proposed approach is based on a specific measure, termed synthetic dispersion measure (SDM), related to the trace of the covariance matrix of the posterior demand estimate conditional upon a set of sensor locations. Under the mild assumption of multivariate normal distribution for the prior demand estimate, the proposed SDM does not depend on the specific values of the counted flows – unknown in the planning stage – but just on the locations of such sensors. From a practical standpoint, a stepwise algorithm is implemented for calculating the proposed measure given a set of link counts, which avoids matrix inversion. In addition, a sequential heuristic algorithm is presented for the application of the proposed NSLP to real contexts. The methodology also allows a formal budget allocation problem to be set between surveys and counts in the planning stage, in order to maximize the overall quality of the demand estimation process.

► Innovative methodology for solving the network sensors location problem. ► Incorporation of the o–d matrix variability using a synthetic dispersion measure. ► Noteworthy particularization to the case of normal prior distribution. ► Implementation of an effective sequential heuristic algorithm. ► Addressing a budget allocation problem between surveys and counts.