WiFiUS: Collaborative Research: Data-Guided Resource Management for Dense Heterogeneous Networks

Purdue University, West Lafayette IN

Investigators

Abstract

The emerging paradigm of dense heterogeneous wireless networks (HetNet), while essential to meeting the explosive growth of wireless traffic, poses unseen challenges to classical cellular network design principles. In traditional cellular systems, spectrum and resource management are often based on the assumption of static and regular topology/traffic patterns. Further, each cell operates with an orthogonal set of resources and requires little inter-cell coordination. However, such traditional approaches are out-dated and inefficient in dense HetNets:the proliferation of small-size cells makes the network topology and traffic characteristics highly irregular and varying; and tighter coordination across cells becomes a necessity when users traverse many small-cells frequently. Thus, there is an acute need to re-examine fundamental ways to design HetNets that can (i) quickly adapt to irregular topology and changing load patterns, and (ii) manage cell coordination at scale with minimal overhead to achieve robust and dependable application-level performance. To address this open challenge, this WiFiUS project develops new framework, architecture and algorithms for adaptive, efficient, and dependable spectrum management and resource allocation in dense heterogeneous cellular networks. First, the project team develops a new hypercell architecture that views a macro-cell and the overlapping small-cells as a single logical entity. This new architecture allows hypercells to acquire a global view of the load/channel/mobility/application patterns across multiple base-cells, thereby enabling more effective and dynamic multi-cell coordination at scale. Furthermore, the team develops a data-guided operational framework that exploits both provider-collected and user-contributed data to address the inherent complexity of dynamic multi-cell coordination. Based on historical data, this approach migrates higher-complexity computations to offline, and thus achieves high spectrum efficiency and user QoE with low-overhead. The US/Finland team brings together a wide range of wireless expertise from physical layer design to network resource management. Having a successful collaboration record, the team carries out joint research activities in the following three aspects: 1) data collection and traffic pattern identification; 2) data-guided spectrum management and base-station cooperation; and 3) hypercell link adaptation, measurement, and resource management. The success of the project has a broad impact on the wireless communications industry by addressing multiple timely and critical challenges that arise from the exponential growth of cellular traffic. The project outcomes will also provide important insights into big-data mining and learning for cellular resource management. The results from this work will be widely disseminated through journal/conference publications, and be incorporated into undergraduate and graduate education endeavors.