Comparing Delay-Constrained ALOHA and CSMA: A Learning-Based Low-Complexity Approximate Approach

Abstract

Supporting delay-constrained traffic becomes more and more critical in multimedia communication systems, tactile Internet, networked control systems, and cyber-physical systems, etc. In delay-constrained traffic, each packet has a hard deadline. When it is not delivered before the hard deadline, it becomes useless and will be removed from the system. This feature is completely different from that of traditional delay-unconstrained traffic and brings new challenge to network protocol design. In this work, we study the widely-used (slotted) ALOHA and CSMA wireless access protocols but under the new delay-constrained setting. Our goal is to compare delay-constrained ALOHA and CSMA for different system settings and thus give network operators guidelines on protocol selection. We use two Markov chains to analyze delay-constrained ALOHA and CSMA, respectively. However, the number of states of Markov chains increases exponentially with respect to the number of users in the network. Therefore, we can only compare the exact performance of delay-constrained ALOHA and CSMA for small-scale networks. To address the curse of dimensionality, we design a single-user parameterized ALOHA (resp. CSMA) system, where the parameters are to be learned to approximate the original multi-user ALOHA (resp. CSMA) system. In addition, our low-complexity approach preserves the Markov-chain structure of the systems and thus enables us to compute some other interested performance metrics such as average delivery time. We use our low-complexity approach to reveal the conditions under which ALOHA (resp. CSMA) outperforms CSMA (resp. ALOHA) in the delay-constrained setting via extensive simulations.

Keywords

• Delay-constrained communications
• ALOHA
• CSMA
• Markov chain
• machine learning