Enhancing the Delay Performance of Dynamic Backpressure Algorithms

Abstract

Abstract-For general multi-hop queueing networks, delay optimal network control has unfortunately been an outstanding problem. The dynamic backpressure (BP) algorithm elegantly achieves throughput optimality, but does not yield good delay performance in general. In this paper, we obtain an asymptotically delay optimal control policy, which resembles the BP algorithm in basing resource allocation and routing on a backpressure calculation, but differs from the BP algorithm in the form of the backpressure calculation employed. The difference suggests a possible reason for the unsatisfactory delay performance of the BP algorithm, i.e., the myopic nature of the BP control. Motivated by this new connection, we introduce a new class of enhanced backpressure-based algorithms which incorporate a general queue-dependent bias function into the backpressure term of the traditional BP algorithm to improve delay performance. These enhanced algorithms exploit queue state information beyond one hop. We prove the throughput optimality and characterize the utility-delay tradeoff of the enhanced algorithms. We further focus on two specific distributed algorithms within this class, which have demonstrably improved delay performance as well as acceptable implementation complexity.Index Terms-dynamic backpressure algorithms, congestion control, delay optimal control, throughput optimal control, dynamic programming, Lyapunov drift.

A Trusted Routing Scheme Using Blockchain and Reinforcement Learning for Wireless Sensor Networks

“…Recently, Gao et al proposed multi-agent Q-learning (QL) aided backpressure routing algorithm named QL-backpressure (BP), where each routing node only needs the local information of the neighbor routing nodes to solve this problem [34]. Their algorithm not only outperforms the BPmin algorithm in delay performance but also contains the excellent characteristics: distributed implementation, low computational complexity, and high-throughput [36]. However, when the malicious nodes appear, the throughput-optimality characteristic will no longer exist.…”

A Multi-Layer Multi-Timescale Network Utility Maximization Framework for the SDN-Based LayBack Architecture Enabling Wireless Backhaul Resource Sharing

“…For instance, Tassiulas and Ephremides [34,35] analyzed Queue length Maximum Weight (QMW) scheduling, which facilitated the subsequent analysis of throughput optimality conditions and related performance guarantees [36,37]. However, QMW scheduling does not guarantee minimal delay [38], which has led to investigations of QMW variations that reduce delays in general multi-hop networks [39] or provide better delay guarantees [40,41]. A common shortcoming of these optimization models is that a centralized optimal scheduler can be impractical, mainly due to scalability problems and signaling delays.…”

Differential back‐pressure routing for single‐queue time‐varying wireless networks

“…In addition to improvements to the weight calculation and backlog measure, some researchers use queueing optimisation to improve delay performance [21–24]. Ji et al introduce a delay‐aware queueing policy to improve the delay performance of the original back‐pressure algorithm [21].…”

“…In this method, suitable virtual‐queue‐based gradients are pre‐built at nodes to reduce the time required to form queue‐backlog‐based gradients. Cui et al [24] use a multi‐hop structure for packet queueing to enhance the knowledge of the backlogs of nodes. The above methods reduce the end‐to‐end delay of packets compared to the original back‐pressure routing algorithm.…”