An Online Delay Efficient Packet Scheduler for M2M Traffic in Industrial Automation

TL;DR

This paper introduces an online, QoS-aware packet scheduler designed for M2M uplink traffic in industrial automation, effectively balancing latency requirements for periodic updates and event-driven data to enhance system utility.

Contribution

The paper proposes a novel online scheduling algorithm that prioritizes event-driven data while ensuring periodic updates meet latency deadlines, improving system utility in industrial M2M networks.

Findings

Outperforms FCFS, EDD, and priority scheduling in simulations.

Effectively balances latency for PU and ED traffic.

Reduces congestion by dropping late PU packets.

Abstract

Some Machine-to-Machine (M2M) communication links particularly those in a industrial automation plant have stringent latency requirements. In this paper, we study the delay-performance for the M2M uplink from the sensors to a Programmable Logic Controller (PLC) in a industrial automation scenario. The uplink traffic can be broadly classified as either Periodic Update (PU) and Event Driven (ED). The PU arrivals from different sensors are periodic, synchronized by the PLC and need to be processed by a prespecified firm latency deadline. On the other hand, the ED arrivals are random, have low-arrival rate, but may need to be processed quickly depending upon the criticality of the application. To accommodate these contrasting Quality-of-Service (QoS) requirements, we model the utility of PU and ED packets using step function and sigmoidal functions of latency respectively. Our goal is to maximize the overall system utility while being proportionally fair to both PU and ED data. To this end, we propose a novel online QoS-aware packet scheduler that gives priority to ED data as long as that results the latency deadline is met for PU data. However as the size of networks increases, we drop the PU packets that fail to meet latency deadline which reduces congestion and improves overall system utility. Using extensive simulations, we compare the performance of our scheme with various scheduling policies such as First-Come-First-Serve (FCFS), Earliest-Due-Date (EDD) and (preemptive) priority. We show that our scheme outperforms the existing schemes for various simulation scenarios.