Relay Selection and Resource Allocation for Ultra-Reliable Uplink Transmission in Smart Factory Scenarios

TL;DR

This paper proposes a relay-assisted transmission protocol for smart factories that optimizes relay selection, resource allocation, and power to ensure ultra-reliable uplink communication within latency constraints, using novel transformation and penalty methods.

Contribution

It introduces a new optimization framework leveraging relative entropy and penalty techniques for relay and resource management in ultra-reliable factory networks.

Findings

The proposed methods effectively reduce total transmit power.

The algorithms demonstrate good convergence properties.

Reliability and relay placement significantly impact power consumption.

Abstract

In this paper, a relay-aided two-phase transmission protocol for the smart factory scenario is proposed. This protocol aims at enabling all robots' ultra-reliable target number of uplink critical data transmission within a latency constraint by jointly optimizing the relay selection, resource block (RB) assignment, and transmit power allocation. Such protocol design is formulated as a mixed-integer and strictly non-convex problem where optimization variables are mutual coupling, which is definitely challenging. Instead of conventional methods designed for solving the problem, we leverage the properties of the relative entropy function to equivalently transform the problem without introducing extra constraints. As the packet error probability requirements of each robot under two possible transmission modes are coupled in one overall reliability constraint, the big-M technique is applied to decouple it into two corresponding reliability constraints. One is for direct transmission mode, and the other is for cooperative transmission mode. Moreover, both non-convex penalty (NCP) and quadratic penalty (QP) approaches are utilized to deal with the binary indicator constraints. Based on such penalty methods, a sequence of penalized approximated convex problems can be iteratively solved for sub-optimal solutions. Numerical results demonstrate the efficiency of such two penalty methods from the perspectives of sub-optimal values of total transmit power and convergence rate. Further, the impacts of reliability, the number and location of relays, the number of robots, the target number of data bits on the total power consumption are analyzed.