Fundamental Limits of Dynamic Interference Management With Flexible Message Assignments and Separate Deep Fading Block Coding

• Published in: IEEE transactions on information theory Vol. 66; no. 2; pp. 1193 - 1212
• Main Authors: Seyfi, Tolunay, Karacora, Yasemin, El Gamal, Aly
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.02.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Block codes; Block erasures; Coding; cooperative zero-forcing; coverage; Degrees of freedom; Encoding; Fading; Fading channels; Interference; message assignment; Network topology; Receivers; shadowing; Topology; Transmitters; Variations
• ISSN: 0018-9448; 1557-9654
• DOI: 10.1109/TIT.2019.2951777

The problem of interference management is considered in the context of a linear interference network that is subject to long term channel fluctuations due to shadow fading. The slow fading model used is one where communication takes place over blocks of time slots and each link in the network is subject independently to block erasure with probability p. It is assumed that each receiver in the network is interested in one unique message, which is made available at M transmitters. For the case where M = 1, the cell association problem is considered, and for M > 1, the problem of setting up the backhaul links for Coordinated Multi-Point (CoMP) transmission is investigated. In both cases, optimal schemes from a Degrees of Freedom (DoF) viewpoint are analyzed for the setting of no erasures, and new schemes are proposed with better average DoF performance at higher probabilities of erasure. Assuming separate coding over different deep fading erasure blocks, the average per user DoF for M = 1 is characterized for every value of p, and optimal message assignments are identified. For M > 1, it is first established that there is no strategy for assigning messages to transmitters in networks that is optimal for all values of p. The optimal cooperative zero-forcing scheme for M = 2 is then identified, and shown to be information-theoretically optimal when the size of the largest subnetwork that contains no erased links is at most five.