C-RAN-Assisted Non-Coherent Grant-Free Random Access Based on Compute-and-Forward

Reliable communication in massive random access scenarios with short packet transmissions requires integration of novel network architecture solutions and multiple access methods. In this paper, we study the performance of an uncoordinated, grant-free random access scheme based on compute-and-forward (CoF) in a dense cloud radio access network (C-RAN) where multiple access points (out-of-band relays or radio units) are connected by low-rate links to a central processing instance. The access points decode linear equations of transmitted messages and forward these to the central processing instance, which solves for the original messages. In the context of the C-RAN architecture, the approach provides efficient utilization of the (limited) fronthaul resources, thus presenting an alternative to conventional C-RAN implementations. The performance of the CoF-based scheme is assessed in a realistic scenario where neither the transmit nor the receive instances have a priori channel state information (CSI). In our approach, pilot signals and data are transmitted in the same message, which meets a strong need for the integration of control and data for short message transmissions. An important aspect that we study is the trade-off between the resource allocation for reliable channel estimation and encoded data transmission given a reliability target.