Cell residence time analysis for performance enhancements of multi-tier heterogeneous networks

Abstract

Owing to the rapid increase of mobile data traffic, the heterogeneous network (HetNet) for increasing the overall network performance and capacity has become a promising technology in fifth-generation (5G) network systems. In the HetNet, small cells, such as picocells, femtocells, and WiFi hotspots are usually positioned by a stochastic process in an unplanned random way. In such complex environments, performance analysis and mathematical modeling of cellular mobile systems have emerged as important issues for network performance enhancements with a diverse set of small cells. Based on the analytical model of cellular mobile networks, mobile service providers can save cost and time, because performance of their new service designs can be analyzed in advance, without wide implementation and deployment. Knowledge of user mobility is a key role of performance analysis and mathematical modeling of cellular networks. In cellular mobile systems, user mobility can be characterized by the cell residence time. Cell residence time is one of the most important system parameters for the performance evaluation of cellular mobile networks and their related applications. Various key network performance metrics, such as network resource occupancy state probabilities, should be calculated through analytical results of the mean cell residence time. However, this is difficult to estimate in 5G heterogeneous networks, due to irregular-shaped multi-tier heterogeneous network topologies and the random movement of mobile users. In order to find an efficient way of calculating the cell residence time for 5G heterogeneous networks, we first examine the well-known conventional formula. Then we propose an enhanced closed-form formula for cell residence time by considering more generalized cellular mobile systems, as well as more generic user mobility models. The proposed closed-form formula can be directly applied to various complex and reliable network scenarios. It can be adapted to multi-tier heterogeneous networks with randomly positioned irregular-shaped small cells, even though their service areas overlap with each other. Our simulation results show that the proposed formula provides excellent estimation accuracy for general random mobility models, including the Levy walk, which is known to realistically reflect human mobility.