Stability Analysis of TDD Networks Revisited: A trade-off between Complexity and Precision

TL;DR

This paper revisits the stability region of TDD networks, considering relaying and link adaptation, and offers a convex polytope characterization along with an approximation method balancing complexity and accuracy.

Contribution

It provides a full characterization of the stability region for TDD networks with relaying and multiple rates, introducing an approximation to reduce computational complexity.

Findings

Exact stability region is a convex polytope with finite vertices.

An approximation reduces complexity with a trade-off in precision.

Numerical results validate the proposed methods.

Abstract

In this paper, we revisit the stability region of a cellular time division duplex (TDD) network. We characterize the queuing stability region of a network model that consists of two types of communications: (i) users communicating with the base station and (ii) users communicating with each other by passing through the base station. When a communication passes through the base station (BS) then a packet cannot be delivered to the destination UE until it is first received by the BS queue from the source UE. Due to the relaying functionality at the BS level, a coupling is created between the queues of the source users and the BS queues. In addition, contrarily to the majority of the existing works where an ON/OFF model of transmission is considered, we assume a link adaptation model (i.e. multiple rate model) where the bit rate of a link depends on its radio conditions. The coupling between the queues as well as the multiple rate model are the main challenges that highly increase the complexity of the stability region characterization. In this paper, we propose a simple approach that permits to overcome these challenges and to provide a full characterization of the exact stability region as a convex polytope with a finite number of vertices. An approximated model is proposed for reducing the computational complexity of the exact stability region. For the multi-user scenario, a trade-off is established between the complexity and the preciseness of the approximated stability region compared to the exact one. Furthermore, numerical results are presented to corroborate our claims.