A Virtualized LoRa Testbed and Experimental Results for Resource Pooling

Traditional network architecture design of Low Power Wide Area Networks (LPWAN) is incapable of dynamically scaling resources based on the served traffic and requires manual procedures for network capacity upgrades. Today's over-provisioning approach based on proprietary hardware (HW) would not be cost and energy efficient to cope with the ever-increasing scale of Internet of Things (IoT) devices in future. The concept of virtualized radio access network (vRAN) by softwarizing and virtualizing communication stacks to a cloud environment is seen as a potential solution to provide web-scale scalability with the benefit of automation and resource pooling, which would significantly increase the resource utilization efficiency. In this work, we developed a LoRa-based testbed to investigate the resource pooling perspective for virtualized physical layer (PHY) functions in a Cloud environment. The testbed is first experimentally validated by comparing the single-cell emulation results and the test results with real LoRa devices. The results show that the testbed can generate the same traffic pattern as real LoRa devices, where two cases show similar resource usages. Then the multi-cell experiments show that one PHY function instance can serve many full-traffic LoRa cells, which showcases a great potential of resource pooling for LoRa networks.