Collaborative Research: SWIFT: LARGE: MAC-on-MAC: A Spectrum Orchestrating Control Plane for Coexisting Wireless Systems
Project runs from 01/01/2021 to 12/31/2023
Wireless systems with different radio access technologies (RATs)nare becoming packed tightly in the space of radio spectrum. The carrier frequency and channel bandwidth of these systems, however, are drastically different across the spectrum domain, e.g., the IEEE 802.11 family that operates at 900 MHz, 2.4/5 GHz, and 60 GHz (mm-wave) bands, the 3GPP 5G family that spans 6-100 GHz, and more. Therefore, radio spectrum, which has long been identified as a scarce resource, will be more crowded and diversified than ever. Swift and effective spectrum sharing requires enhanced capability and increased intelligence at the wireless devices, from innovative transmitter and receiver technologies at the physical layer, to multi-band spectrum sensing across physical and MAC layer; from mapping spectrum slices for each access request, to radical modification of medium access control (MAC) protocols. Altogether, there is a need to support a myriad of heterogeneous devices that run diverse communication standards over different frequency bands to achieve efficient spectrum and energy utilization in such huge, dynamic and disparate systems.
This project mitigates the incoherent and disassociated frequency bands of the multi-RAT coexisting environments by exploiting the potential of cross-layer design from the circuits of transmitter and receiver, to the MAC layer with an innovative MAC-on-MAC spectrum control plane. This project has four major thrusts: (i) design of innovative transmitter and receiver techniques for energy-efficient multi-band spectrum monitoring, by using custom single-chip ultra-broadband (1-8 GHz) bio-inspired spectrum sensors to create a multi-band sensing solution that includes both sub-6 GHz and emerging mm-wave bands; (ii) methods to acquire complete spectrum occupancy information with a limited number of measurements from the proposed multi-band spectrum sensing circuits by exploiting matrix completion for low-cost, accurate and scalable monitoring through a learning-based sequential scheme; (iii) algorithms of obtaining a sorted ranking of paths that map a data flow to a range of spectrum slips by modeling spectrum-flow-RAT domains, finding the set of unoccupied spectrum slices, sorting out multiple recommendations from neighboring SAP control entities; (iv) creating MAC-Flow, a flow based network protocol that is capable of handling networking demands from both upper and lower layer, for both traditional user frames and for MAC-on-MAC controller processed frames for system-level performance evaluation