A real-time origin-destination matrix updating algorithm for on-line applications

This study develops an improved origin-destination matrix updating algorithm based on the existing multiplicative algebraic reconstruction technique with the entropy-maximizing model. The improvement of this algorithm in numerical stability and convergence speed is obtained by incorporating a normalization technique and a diagonal searching strategy. Two sets of numerical examples are also reported to show the experimental performance of the proposed algorithm in comparison with two other prevailing algorithms. The first set of example problems are based on randomly generated data scenarios with various dimensions, while the other set is based on the microscopic simulation model, CORSIM. All numerical results have demonstrated that the proposed new algorithm can substantially save computation time compared to the two other algorithms, and thus holds much promise for efficient on-line application in advanced traffic management systems.