Achieving Robust Resource Orchestration for Highly Dense Heterogeneous IoT Systems

The proliferation of the Internet of Things (IoT) has ushered in a wide array of emerging applications. Current mainstream IoT protocols for supporting the above applications, such as Wi-Fi, ZigBee, and Bluetooth, heavily overlap on the 2.4 GHz bands. When deploying those heterogeneous IoT devices with different wireless protocols in a limited geographic area, e.g., manufacturing warehouse and clinic rooms, inevitable packet collisions will occur due to their spectrum overlapping. Those unpredictable collisions will ultimately degrade network performance, mainly due to the lack of coordination across coexisting protocols. This paper revisits the classic resource orchestration problem in a practical wireless coexistence scenario with a dense indoor IoT deployment. We propose to leverage multi-protocol gateways, e.g., Amazon Echo, Google Nest Hub, and Samsung SmartThing Station, to develop a Multi-Agent Reinforcement Learning (MARL) framework, which jointly considers channel status and contextual information for orchestrating limited resources. Based on the protocol heterogeneity and diverse transmission requests, we design a novel resource pool to achieve fine-grained management of available resources, by which gateways can collaboratively decide the system-level optimal strategy. The proposed design will also feature a cascaded RL model to determine a sequential decision for best utilizing available resources. Based on extensive real-world experiments conducted on a Software-Defined Radio (SDR) platform with up to 33 IoT devices, our proposed framework achieves more than 2.19X in throughput. It reduces 69.07% of delay compared with current random-accessed mechanisms.