Coding of Multi-Source Information Streams with Age of Information Requirements

Haoyuan Pan,Soung Chang Liew,Jiaxin Liang,Victor C. M. Leung,Jianqiang Li

DOI: https://doi.org/10.1109/jsac.2021.3065079

IF: 16.4

2021-01-01

IEEE Journal on Selected Areas in Communications

Abstract:This article puts forth a new channel coding paradigm for multi-source information streams with Age of Information (AoI) requirements. The recently introduced AoI metric characterizes the freshness of information, defined as the time elapsed since the generation of the last successfully received update. We study a setup in which a large number of sensors want to send update information to a common monitor with the help of aggregators. Specifically, an aggregator collects update packets from sensors and forwards them to the monitor. Conventional block codes (such as LDPC codes) that encode and decode each update packet separately do not perform well in such an information aggregation and update scenario. When update packets suffer from packet loss, we show that block codes lead to high instantaneous AoI because a sensor waits for a long time for the next update opportunity. This article investigates stream-based codes to tackle this problem. A distinguishing feature of stream-based codes is the joint encoding of update packets from different sensors, and a series of coded packets are sent continuously like a stream. Different update packets are then jointly decoded using multiple coded packets from the stream. A key challenge with AoI requirements is the joint design of error corrections of old packets and fast decodings of new packets. We design a practical encoding-decoding scheme and a sliding decoding window mechanism to control the decoding complexity. We evaluate two AoI metrics, average AoI and bounded AoI. In particular, bounded AoI corresponds to an AoI threshold that the instantaneous AoI is below a large percentage of the time. Experimental results on software-defined radio show that stream-based codes significantly outperform block codes in both average AoI and bounded AoI under varying channel conditions. Overall, stream-based codes provide a viable channel coding solution to multi-source information streams with timely update requirements.