Research

PI: Prof. Kwang-Cheng Chen   
Intel Champion: Dr. Yen-KuangChen

The goal of project CIFI and SNis to study effective communications in machine swarm of cloud based environments as Figure shown. More precisely, CIFI develops spectral efficient communication and networking among numerous machines to enable cyber-physical systems, using information fusion, inference, and cooperative multi-hop networking, under possible spectrum sharing environments.

Under this challenging scenario, we explore fundamental technology including

• Spectrum sharing multi-hop cooperative networking: Underlarge number of machines, spectrum sharing would be needed. With priority among heterogeneous systems/networks, we study cognitive radio networks. Without priority among heterogeneous systems/networks, we study spectrum sharing heterogeneous networking.
• Using information fusion and inference to establish a spectrum map to enable multi-hop networking in machine swarm: To avoid communication overhead and to determine the optimal routes, we need the knowledge of radio channel conditions that can be provided by establishing a spectrum map. The spectrum map indicates the average channel qualities (radio power or interference level) at different locations. Several innovative approaches have been considered: compressive sensing, tomography, and heterogeneous information fusion and inference(HIFI).
• Traffic reduction: To analyze how much traffic can be reduced by fusing data or inferring information from multiple machines at various points along potential M2M transmission routes that are determined in Target A. The“coding” part is to reduce the rate of information flow at each node by considering the correlation among information flows from other nodes within transmission range. The“forward” part is to choose the optimal next hop (and thus the optimal route) such that the aggregate rate of information flows is minimized.
• Socially enabled heterogeneous networking architecture and algorithms: To resolve the scalability of ad-hoc networking, we investigate the possibility to achieve multi-hop networking of large number of machines under delay constraints and QoS requirements. Proper networking algorithms with above goals are expected as the final project outputs.