Distributed Estimation of Graph Eigenspectra in Mobile Communication Networks

For large-scale mobile communication networks, their graph eigenspectra directly affect the performance of congestion control, flow control, quality of service and so on. At present, a distributed estimation is considered to be the most suitable way for signal processing as for infrastructure-free networks. However, due to the structural dynamics of mobile communication networks, it is hard for the nodes to perceive the current state parameters of an entire network, especially at the stage of nonsteady-state of network dynamic systems. This paper accordingly proposes a distributed method to estimate the graph eigenspectra of mobile communication networks on the basis of the singular value decomposition theory. Second, authors establish a Kalman Filtering model for real-time tracking the estimation of graph eigenspectra. Finally, through the theoretical analysis and numerical simulations, it is shown that our proposed estimation method can satisfy the accuracy requirements for the network reconstruction.