A Tractable Analysis Model of Information Freshness for Mobile Edge Computing Assisted IoT System With Layer-Coded HARQ

This work focuses on an Internet of Things (IoT) status update system with the assistance of mobile edge computing (MEC) under short packet communications. The edge server includes a successive interference canceller (SIC) and a buffer pool. To simultaneously improve both information timeliness and throughput, a layer-coded hybrid automatic repeat request (L-HARQ) protocol is proposed especially considering backtrack decoding (BD) delay. In the MEC-assisted L-HARQ IoT status update system, the source successively transmits mixing updates to an access point (AP). The mixing update consists of the new generation and the part of the previously failed one. All the undecoded mixing packets are delivered to the edge server. Once a successful feedforward decoding (FD) occurs, the edge server attempts to recover undecoded mixing packets until a BD failure occurs or all buffers empty. The successful BD results are also delivered to the IR and the obtained prior information is fed back to the edge server for the next BD. By employing the stochastic hybrid system (SHS) model, the average age of informations (AoIs) are derived under circle-shift preemption (CS-P) and fixed preemption (F-P) policies. Considering the fact that the number of buffers is limited and the successful FD and BD do not always cause the reduction of system AoI, this work proposes a non-binary age evolution model as well as the state simplification mothed in SHS. The presented numerical results show that the CS-P and F-P policies obtain the enhanced information freshness compared with the non-preemptive blocking policy, and the CS-P policy outperforms the F-P one in terms of information freshness. The average AoI greatly depends on BD depth. When it is small, the average AoI is small. However, the two policies have the same average throughput that mainly depends on the size of buffer pool and the success probabilities of both FD and BD.