Abstract
The current development of the Internet of Things (IoT) suffers from problems such as scattered deployments, different structures, and inefficient interconnections of networks. Therefore, the current IoT architecture cannot support complex IoT applications that coordinate and reuse cross-industry resources. This paper proposes an application-driven ternary (i.e., data, resources, and services) divide-and-conquer architecture of heterogeneous IoT. The architecture supports decoupling and reconstructing heterogeneous services on demand based on the characteristics of upper-layer applications and performing adaptive management and dynamic scheduling of heterogeneous resources. To achieve the interconnection of different types of IoT systems, our architecture logically virtualizes multiple IoT systems into a complete network system for supporting data interaction, resource reusing, and service collaboration. To validate the effectiveness of our architecture, we apply it to the environmental monitoring scenario, where robust Chinese Remainder Theorem (RCRT) is leveraged to implement the data aggregation between sensing devices and the IoT gateway. The experimental results show that our architecture can meet the requirements of the scenario and enable up to 10 billion devices to be connected theoretically.