[Paper Review] On the Throughput Maximization in Dencentralized Wireless Networks

Jamshid Abouei, Alireza Bayesteh+2 more|ArXiv.org|Oct 14, 2008

Cooperative Communication and Network Coding37 references3 citations

TL;DR

This paper proposes a distributed, non-iterative power allocation strategy in decentralized wireless networks with K links, using orthogonal subchannels (M clusters) and a shadow-fading model. It proves that maximum network throughput—defined as average sum-rate or guaranteed sum-rate—scales as Θ(log K) and is maximized when M=1, with an on-off power allocation scheme under strong interference.

ABSTRACT

A distributed single-hop wireless network with $K$ links is considered, where the links are partitioned into a fixed number ($M$) of clusters each operating in a subchannel with bandwidth $\frac{W}{M}$. The subchannels are assumed to be orthogonal to each other. A general shadow-fading model, described by parameters $(α,\varpi)$, is considered where $α$ denotes the probability of shadowing and $\varpi$ ($\varpi \leq 1$) represents the average cross-link gains. The main goal of this paper is to find the maximum network throughput in the asymptotic regime of $K o \infty$, which is achieved by: i) proposing a distributed and non-iterative power allocation strategy, where the objective of each user is to maximize its best estimate (based on its local information, i.e., direct channel gain) of the average network throughput, and ii) choosing the optimum value for $M$. In the first part of the paper, the network hroughput is defined as the extit{average sum-rate} of the network, which is shown to scale as $Θ(\log K)$. Moreover, it is proved that in the strong interference scenario, the optimum power allocation strategy for each user is a threshold-based on-off scheme. In the second part, the network throughput is defined as the extit{guaranteed sum-rate}, when the outage probability approaches zero. In this scenario, it is demonstrated that the on-off power allocation scheme maximizes the throughput, which scales as $\frac{W}{α\varpi} \log K$. Moreover, the optimum spectrum sharing for maximizing the average sum-rate and the guaranteed sum-rate is achieved at M=1.

Motivation & Objective

• To maximize network throughput in large-scale decentralized wireless networks with K links and limited spectrum.
• To address the impracticality of centralized resource allocation in large networks by developing a distributed, non-iterative power control strategy.
• To determine the optimal number of subchannels M that maximizes average and guaranteed sum-rate under shadow fading.
• To establish that throughput scales logarithmically with K, and that M=1 achieves the maximum throughput in both average and guaranteed sum-rate regimes.

Proposed method

• Models a wireless network with K single-hop links partitioned into M clusters, each using an orthogonal subchannel of bandwidth W/M.
• Uses a general shadow-fading model with parameters (α, ϖ), where α is the shadowing probability and ϖ is the average cross-link gain.
• Proposes a distributed, non-iterative power allocation strategy where each user sets power based only on its local direct channel gain to maximize its estimate of network throughput.
• Analyzes two throughput metrics: (1) average sum-rate, and (2) guaranteed sum-rate under outage probability approaching zero.
• Derives asymptotic throughput scaling laws using stochastic geometry and extreme value analysis, particularly for K→∞.
• Uses the on-off power allocation scheme in strong interference scenarios, proving it maximizes throughput under both throughput definitions.

Experimental results

Research questions

• RQ1 What is the optimal number of subchannels M that maximizes network throughput in a decentralized wireless network with K links?
• RQ2 How does the network throughput scale asymptotically as K→∞ under a general shadow-fading model?
• RQ3 Can a distributed, non-iterative power allocation strategy based on local channel knowledge achieve optimal or near-optimal throughput?
• RQ4 Is there a fundamental difference in optimal spectrum sharing (M) between average sum-rate and guaranteed sum-rate regimes?
• RQ5 Does the on-off power allocation scheme maximize throughput under strong interference, and if so, why?

Key findings

• The average sum-rate of the network scales as Θ(log K) in the asymptotic regime K→∞, regardless of M.
• For the guaranteed sum-rate, the maximum throughput scales as (W/(αϖ)) log K, and is achieved using the on-off power allocation scheme.
• The optimal spectrum sharing for both average and guaranteed sum-rate is achieved at M=1, meaning full bandwidth should be used by a single cluster.
• The on-off power allocation strategy maximizes throughput in strong interference scenarios, as proven via extremum analysis of the throughput function.
• For the case of no shadowing (α=0), the average sum-rate is a monotonically decreasing function of M, confirming that M=1 maximizes throughput.
• Asymptotic analysis shows that when M=K (i.e., each link uses a separate subchannel), the average sum-rate approaches W(log K − log(N₀W) − γ), confirming logarithmic scaling.