ML based velocity estimator via gamma distributed handover counts in HetNets

Abstract

In this paper, we present a maximum likelihood (ML) based approach to estimate velocity of mobile users in Heterogeneous Networks (HetNets). HetNets by architecture are hierarchal combination of randomly deployed base stations (BSs) with varied transmit power levels and hence have non-uniform coverage area. Further, in order to improve network capacity, BS density in HetNets is much higher in comparison to traditional cellular networks. The increased BS densification in HetNets results in frequent handovers which if not managed properly may leads to service failures. One of the fundamental challenge in effective handover management is to accurately estimate the velocity of mobile users. Thus, we propose a velocity estimation strategy based on handover count samples which is in accordance with Release-8 of LTE specifications. We analyze the handover statistics by modeling BS location via stochastic geometry and coverage area by Poisson-Voronoi tessellation. The probability mass function (PMF) of handover count is approximated via Gamma distribution as it has very small approximation error compared to Gaussian distribution. Using the approximated PMF, we first derive maximum likelihood (ML) base velocity estimator for the respective mobile user in the network. In addition, we also derive the Cramer-Rao lower bound (CRLB). We validate our proposed estimation approach via numerical results in which we observe tight closeness between asymptotic variance of estimated velocity and CRLB. Our results also demonstrate that velocity estimation error decreases individually with increase in BS density and time duration specified for handover count measurements.