Engineering Edge-Cloud Offloading of Big Data for Channel Modelling in THz-range Communications

Channel estimation in mmWave and THz-range wireless communications (producing Gb/Tb-range of data) is critical to configuring system parameters related to transmission signal quality, and yet it remains a daunting challenge both in software and hardware. Current methods of channel estimations, be it modeling- or data-based (machine learning (ML)), - use and create big data. This in turn requires a large amount of computational resources, read operations to prove if there is some predefined channel configurations, e.g., QoS requirements, in the database, as well as write operations to store the new combinations of QoS parameters in the database. Especially the ML-based approach requires high computational and storage resources, low latency and a higher hardware flexibility. In this paper, we engineer and study the offloading of the above operations to edge and cloud computing systems to understand the suitability of edge and cloud computing to provide rapid response with channel and link configuration parameters on the example of THz channel modeling. We evaluate the performance of the engineered system when the computational and storage resources are orchestrated based on: 1) monolithic architecture, 2) microservices architectures, both in edge-cloud based approach. For microservices approach, we engineer both Docker Swarm and Kubernetes systems. The measurements show a great promise of edge computing and microservices that can quickly respond to properly configure parameters and improve transmission distance and signal quality with ultra-high speed wireless communications.