Scheduling in densified networks: Algorithms and performance

With increasing data demand, wireless networks are evolving to a hierarchical architecture where coverage is provided by both wide-area base-stations (BS) and dense deployments of short-range access nodes (AN) (e.g., small cells). The dense scale and mobility of users provide new challenges for scheduling: (i) High flux in mobile-to-AN associations, where mobile nodes quickly change associations with access nodes (time-scale of seconds) due to their small footprint, and (ii) multi-point connectivity, where mobile nodes are simultaneously connected to several access nodes at any time. We study such a densified scenario with multi-channel wireless links (e.g., multi-channel OFDM) between nodes (BS/AN/mobile). We first show that traditional algorithms that forward each packet at most once, either to a single access node or a mobile user, do not have good delay performance. We argue that the fast association dynamics between access nodes and mobile users necessitate a multi-point relaying strategy, where multiple access nodes have duplicate copies of the data, and coordinate to deliver data to the mobile user. Surprisingly, despite data replication and no coordination between ANs, we show that our algorithm (a distributed scheduler - DIST) can approximately stabilize the system in large-scale instantiations of this setting, and further, performs well from a queue-length/delay perspective (shown via large deviation bounds).