DNA-GA: A Tractable Approach for Performance Analysis of Uplink Cellular Networks

In this paper, we propose a tractable semi-analytical approach for network performance analysis of uplink (UL) cellular networks, which is based on a deterministic network analysis using a Gaussian approximation (DNA-GA). The key contribution of this work is to investigate the UL signal-tointerference ratio (SIR) performance using the DNA-GA analysis. In particular, the SIR is modeled as a ratio of two random variables (RVs), representing the signal power and the aggregate interference power, respectively. The signal power is further characterized by a product of two RVs, i.e., a lognormal RV and a RV with an arbitrary distribution. The former RV comes from a common assumption of lognormal shadow fading, and the latter one takes the rest of random factors into account, such as random user positions, arbitrary types of multi-path fading, etc. The aggregate interference power is approximated by a RV with a power lognormal distribution. The proposed DNA-GA analysis has a few desirable features: (i) it naturally considers lognormal shadow fading; (ii) it can treat arbitrary shape and/or size of cell coverage areas; (iii) it can handle non-uniform user distributions; (iv) it can cope with any type of multi-path fading; (v) it can be applied to multi-antenna base stations (BSs). These features make the DNA-GA analysis very useful for network performance analysis of the 5th generation (5G) systems with general cell deployment and user distribution.