Joint Lossy Compression and Power Allocation in Low Latency Wireless Communications for IIoT: A Cross-Layer Approach

Shaoling Hu,Wei Chen

DOI: https://doi.org/10.1109/tcomm.2021.3077948

IF: 6.166

2021-01-01

IEEE Transactions on Communications

Abstract:Low-latency communication is expected to play a key role in the Industrial Internet of Things (IIoT). Although there has been considerable effort on reducing latency, few attention has been focused on the low-latency transmission of distortion-tolerant data, e.g. IIoT’s analog samples, over fading channels. In this paper, we present a cross-layer approach to jointly adapt the transmission power, rate, and compression ratio based on the instantaneous buffer and channel states. In particular, to minimize the average delay under both average power and distortion constraints, we formulate a Constrained Markov Decision Process (CMDP) with multi-dimensional state and action spaces. A novel solution is then presented by judiciously decomposing the multi-dimensional CMDP into a deterministic hierarchical optimization consisting of a linear programming and a convex optimization problem. Furthermore, we show the optimality of a threshold-based scheduling policy that highly reduces the complexity and present an optimal delay-power-distortion tradeoff that characterizes a fundamental performance tradeoff between the physical, network, and application layers. The joint lossy compression and power allocation scheme realizes a deep cross-layer optimization of source coding, queuing control, and wireless transmission, which outperforms the traditional cross-layer design consisting of only two layers, not to mention layered protocols.