Throughput-Delay-Reliability Tradeoff with ARQ in Wireless Ad Hoc Networks

Rahul Vaze

Abstract

In wireless ad hoc networks, the interplay between throughput, delay, and reliability is a fundamental constraint on performance. This research derives the delay-reliability (D-R) and throughput-delay-reliability (T-D-R) tradeoffs for both single-hop and multi-hop transmissions using Automatic Repeat Request (ARQ). By modeling the delay constraint as a maximum number of end-to-end retransmissions $D$, and defining reliability as the probability of success within that budget, the study characterizes throughput via transmission capacity. A key contribution is the optimization of retransmission allocation across hops and the demonstration that single-hop transmission is optimal for maximizing throughput in sparse network regimes.

Introduction

Wireless ad hoc networks operate without centralized infrastructure, making interference management a primary challenge. To ensure reliable communication, these networks often employ Automatic Repeat Request (ARQ) protocols, where packets are retransmitted until an acknowledgment is received or a retry limit is reached. While ARQ enhances reliability, it inherently introduces delay and consumes network resources, impacting overall throughput.

This paper, authored by Rahul Vaze, provides a rigorous mathematical analysis of the Throughput-Delay-Reliability (T-D-R) tradeoff. Unlike previous models that assume infinite retry budgets or independent interference between time slots, this work addresses the "temporally correlated" nature of interference in networks with static nodes.

The Challenge of Correlated Interference

In a decentralized network, interference is the primary bottleneck. Because nodes in ad hoc networks are often static or slow-moving, the interference environment is temporally correlated. If a transmission fails because a nearby neighbor is active, a rapid retransmission is likely to fail again because that neighbor is likely still active. This "bad neighborhood" effect means that reliability does not improve linearly with the number of retries as it might in a memoryless channel.

The research models the locations of wireless nodes as a Homogeneous Poisson Point Process (PPP) and utilizes Slotted ALOHA for random access. To handle the analytical intractability of joint interference distributions, the author employs the FKG inequality to derive tight upper and lower bounds on the probability of success.

Methodology and Metrics

The study evaluates performance using Transmission Capacity (TC), defined as the maximum allowable density of transmitting nodes that can satisfy a specific rate and outage probability constraint. This metric is then multiplied by the transmission rate and the success probability to characterize the network's effective throughput.

The delay constraint is modeled by an end-to-end budget $D$. A packet is considered successful if it is decoded within $D$ retransmissions. The paper investigates how to best allocate these $D$ retries across $N$ hops in a multi-hop path.

Key Insights

1. Optimal Retransmission Allocation

For a multi-hop path consisting of equal-length hops, the research demonstrates that the optimal strategy to maximize the lower bound of transmission capacity is to distribute the retransmission budget equally. Specifically, if $D$ is the total budget and $N$ is the number of hops, the optimal allocation is $D_n = D/N$ retries per hop.

2. The Multi-hop Myth in Sparse Networks

A significant finding of the paper challenges the conventional wisdom regarding multi-hop routing. In "sparse" regimes—where node density is low—the benefits of shorter hops (which reduce path loss) are often negated by the cumulative delay and the increased number of active nodes required for routing. The study proves that for sparse networks ($\lambda \to 0$), a single long hop ($N=1$) is actually the optimal configuration for maximizing the throughput lower bound under a strict delay constraint.

3. Reliability Scaling

The paper provides exact expressions for success probability ($P_s$). For single-hop transmission, success probability is shown to increase as $1 - x^{D+1}$, where $x < 1$ is a constant determined by network density and the Signal-to-Interference Ratio (SIR) threshold. Numerical simulations confirm that the derived lower bounds are remarkably tight, providing a reliable tool for network design.

Conclusion

By accounting for finite delay budgets and temporally correlated interference, this work provides a more realistic framework for understanding ad hoc network performance. The results suggest that in many practical scenarios, particularly sparse deployments, simplifying the network architecture to fewer hops can yield better throughput-delay-reliability tradeoffs than complex multi-hop routing. Future research may expand these findings by considering node mobility, which could decorrelate interference and potentially shift the balance back toward multi-hop strategies.

Summary updated

Abstract

Delay-reliability (D-R), and throughput-delay-reliability (T-D-R) tradeoffs in an ad hoc network are derived for single hop and multi-hop transmission with automatic repeat request (ARQ) on each hop. The delay constraint is modeled by assuming that each packet is allowed at most $D$ retransmissions end-to-end, and the reliability is defined as the probability that the packet is successfully decoded in at most $D$ retransmissions. The throughput of the ad hoc network is characterized by the transmission capacity, which is defined to be the maximum allowable density of transmitting nodes satisfying a per transmitter receiver rate, and an outage probability constraint, multiplied with the rate of transmission and the success probability. Given an end-to-end retransmission constraint of $D$, the optimal allocation of the number of retransmissions allowed at each hop is derived that maximizes a lower bound on the transmission capacity. Optimizing over the number of hops, single hop transmission is shown to be optimal for maximizing a lower bound on the transmission capacity in the sparse network regime.
BibTeX
@article{vaze2011throughput,
  author = {Rahul Vaze},
  title = {{Throughput-Delay-Reliability Tradeoff with ARQ in Wireless Ad Hoc Networks}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2011},
  url = {https://openreview.net/forum?id=gV8iy5JYAG},
}
Figure