Throughput of Cellular Uplink with Dynamic User Activity and Cooperative Base-Stations

Oren Somekh, Osvaldo Simeone, H. Vincent Poor, Shlomo ShamaiarXiv 2009 · DOI · OpenReview

Abstract

The throughput of a linear cellular uplink with a random number of users, different power control schemes, and cooperative base stations is considered in the large system limit where the number of cells is large for non-fading Gaussian channels. The analysis is facilitated by establishing an analogy between the cellular channel per-cell throughput with joint multi-cell processing (MCP), and the rate of a deterministic inter-symbol interference (ISI) channel with flat fading. It is shown that, under certain conditions, the dynamics of cellular systems (i.e., a random number of users coupled with a given power control scheme) can be interpreted, as far as the uplink throughput is concerned, as the flat fading process of the equivalent ISI channel. The results are used to demonstrate the benefits of MCP over the conventional single cell processing approach as a function of various system parameters in the presence of random user activity.

Introduction

In traditional cellular networks, each base station (BS) independently manages users within its cell. This approach, known as Single-Cell Processing (SCP), treats signals from neighboring cells as interference, which creates an "interference floor" that limits system capacity regardless of transmit power. Modern network architectures propose Joint Multi-Cell Processing (MCP), where base stations cooperate via a backhaul network to jointly decode signals. This transforms inter-cell interference into useful information, potentially offering significant capacity gains.

However, a major challenge in modeling these systems is the dynamic nature of user activity. In real-world scenarios, the number of active users in a cell fluctuates randomly as sessions begin and end. This paper addresses the mathematical complexity of analyzing MCP performance under such dynamic conditions.

System Model and Methodology

The authors utilize a linear cellular model, often referred to as the Wyner model, where cells are arranged in a line and interference is restricted to immediate neighbors. This simplification allows for a tractable mathematical analysis while preserving the fundamental physics of cellular interference.

Figure 1: Linear cellular uplink model with cooperative base stations and a central processor.

A core contribution of this work is the "ISI Analogy." The researchers prove that a cellular uplink with multiple random users is mathematically equivalent to a single-user Inter-Symbol Interference (ISI) channel experiencing flat fading. In this framework, the random fluctuations in user activity are treated as a fading process.

The study evaluates three power control strategies:

  1. No Power Control (NPC): Users transmit at a fixed power level regardless of activity.
  2. Adaptive Power Control (APC): Total cell power is shared equally among currently active users.
  3. Cognitive Power Control (CPC): A theoretical upper bound where inactive users' power is utilized to assist active users.

Key Results and Performance Analysis

The analytical derivations were validated using extensive Monte-Carlo simulations, showing a near-perfect match. A primary finding is the clear superiority of MCP over SCP, especially at high power levels.

Figure 3: Per-cell throughput vs. total-cell power for various processing and power control schemes.

As shown in Figure 3, SCP throughput (dashed lines) eventually plateaus due to the interference floor. In contrast, MCP throughput (solid lines) continues to grow logarithmically with power. Furthermore, Adaptive Power Control (APC) is shown to effectively mitigate the capacity loss typically associated with random user inactivity, bringing performance close to the "always active" scenario.

Another counter-intuitive insight is that in MCP systems, increasing the inter-cell interference factor ($\alpha$) can actually improve throughput. This occurs because cooperative base stations can exploit the stronger redundant signals received from neighboring cells, whereas in SCP, higher $\alpha$ always degrades performance.

Conclusion and Future Work

By establishing a link between dynamic cellular activity and fading ISI channels, this research provides a powerful tool for predicting the performance of next-generation cooperative networks. While the current study focuses on a 1D linear topology with ideal backhaul, it lays the groundwork for analyzing more complex 2D hexagonal grids and non-ideal network conditions. Future research will explore the impact of synchronization errors and limited backhaul capacity on these cooperative gains.

Summary updated

Abstract

The throughput of a linear cellular uplink with a random number of users, different power control schemes, and cooperative base stations is considered in the large system limit where the number of cells is large for non fading Gaussian channels. The analysis is facilitated by establishing an analogy between the cellular channel per-cell throughput with joint multi-cell processing (MCP), and the rate of a deterministic inter-symbol interference (ISI) channel with flat fading. It is shown that, under certain conditions, the dynamics of cellular systems (i.e., a random number of users coupled with a given power control scheme) can be interpreted, as far as the uplink throughput is concerned, as the flat fading process of the equivalent ISI channel. The results are used to demonstrate the benefits of MCP over the conventional single cell processing approach as a function of various system parameters in the presence of random user activity.
BibTeX
@misc{somekh2009throughput,
  author = {Oren Somekh and Osvaldo Simeone and H. Vincent Poor and Shlomo Shamai},
  title = {{Throughput of Cellular Uplink with Dynamic User Activity and Cooperative Base-Stations}},
  year = {2009},
  eprint = {0908.3098},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/0908.3098},
}
Figure