WiFiUS: Collaborative Research: Scalable Edge Architecture for Massive Location-Aware Heterogeneous IoT Systems

Columbia University, New York NY

Investigators

Abstract

The "Scalable Edge Architecture for Massive Location-Aware Heterogeneous IoT Systems" project addresses essential research problems for developing network design and Internet of Things (IoT) systems in both high bandwidth and low bandwidth environments. By defining an efficient security and scalability-enhancing edge architecture that moves data processing close to users, it minimizes data transfer latencies and overhead in the network. The fusion of sensor data from different sources can significantly improve the efficiency of many existing systems. In some areas, massive bandwidth is about to become available to cellular and WiFi networks in the millimeter wave bands. The Federal Government recently opened up 28 GHz at frequencies above 24 GHz, which promises to revolutionize wireless systems and enable IoT applications never before conceived, particularly in dense urban areas. Smart traffic and connected (autonomous) cars are an example application area enabled by the new bandwidth, which requires combining these different approaches in the architectural design of large-scale IoT systems. Integrating IoT with edge/fog computing, millimeter wave (mmWave) technologies and distributed processing enables optimizing the system level performance considering system capacity, reduced network bandwidth for data and control traffic, increased system level programmability and automation, accurate location-awareness, virtualization, low latency, scalability, and enhanced security and privacy. One of the novel aspects of our proposed system is that it transitions seamlessly from an emulated and simulated environment to actual production deployment, and mixtures of these modes, facilitating robust and reliable IoT systems at scale. The project provides contributions in several essential areas of IoT network architectures and security: 1) minimizing the need of manual configuration in large-scale IoT networks; 2) scalable authentication and key management systems; 3) efficient distributed IoT architecture to perform complex and delay-sensitive tasks, with rapid deployment and prototyping; 4) secure interoperability between heterogeneous IoT devices; and 5) positioning and capacity optimization based on mmWave communications, considering especially smart traffic and vehicle-to-vehicle communications. The project proposes a multi-layered approach for scaling IoT systems in simulation and emulation, allowing to combine physical systems with emulated systems, incorporating new mmWave RF and network models, network emulation, virtual systems, models of the physical world and user interfaces. The emulation system allows the team to explore mobile edge and fog computing to enhance efficiency, reduce control-loop delays and assure privacy of sensitive data. A new authentication model and naming system allows scaling for deployment and programming. The prototype open-source IoT emulator, along with the mmWave channel models, developed in the project will allow industrial IoT system developers to more rapidly and reliably develop new IoT systems. The authentication and naming components will be submitted for possible standardization. The new channel models are likely to inform spectrum allocation decisions for mmWave bands by national regulators.