The Internet of Things offers a grand vision. Finding solutions to materialize this vision has become a long-term research goal for a team consisting of professors from the Massachusetts Institute of Technology (MIT), the Chinese University of Hong Kong (CUHK) and the Hong Kong University of Science and Technology (HKUST).
The team members, professors Moe Win, Guy Bresler, Sanjoy Mitter and Yury Polyanskiy from MIT, professors Wing Shing Wong and Chandra Nair from CUHK, as well as professors Li Qiu and Ling Shi from HKUST, come from different engineering research fields, ranging from wireless communication, information theory, networked control and optimization. Yet, they share the same interest in trying to devise a platform for effective management of largenumber of IoT devices in an indoor environment. One potential application for such a platform is to provide precise tracking and fast control of mobile robots in a hospitals.
From a macroscopic architectural viewpoint, research challenges to achieving this goal can be grouped into the three following areas:
Objective 1: Ubiquitous network localization
It is essential to enable location awareness for IoT devices because location is a crucial element to enabling many new applications (e.g. Uber). However, solutions such as cooperative localization, in which devices can communicate with their neighbors, may not be possible in many applications.
This need, and lack of an obvious answer, opens the door for newer approaches to exploit localization services inherent across the broad network operating environment. To do this, the team is proposing:
• Exploitation of Environmental Knowledge: The utilization of this knowledge to develop efficient, robust information fusion techniques.
• Harnessing Signals of Opportunities: Developing new space-time processing methods to effectively exploit them.
The team is experienced in developing theory, algorithms and hardware platforms that can deliver network localization algorithms and achieve location awareness through environmental triggers. In turn, they can simulate studies to evaluate these algorithms.
Objective 2: Networking of IoT devices
This proposed objective is to enable innumerable M2M communications for diverse, sporadic traffic patterns. But the challenge lies in how the high density and variety of IoT devices precludes the efficient use of traditional networking protocols. This necessitates new communication protocols within the network operating across multiple access layers.
The team is proposing the use of network topology analysis to determine the fundamental limits on communication flow depending on topology in combination with multiple access protocols that can tolerate low-synchronization fidelity. Machine learning will also be used to adapt to various communication spectrums.
Though the study’s primary application was discussed in transportation scenarios, the fundamental IoT technology holds potential for use in any number of IoT communications applications, with strong potential in industrial sectors.
Additionally, the team is capable of developing theory and algorithms for multiple access protocols for ad-hoc networks, low-latency communication and analysis of locally interacting systems.
Objective 3: Networked control and optimization
The final objective is to enable distributed, time-critical control of IoT devices, interacting through busy communication networks.
The challenge lies in how existing network designs and protocols aren’t suitable for time-critical IoT applications. Instead, there needs to be integrated models to enable the design of better algorithms for networked control.
The team proposes distributed decision-making and dynamic control, for nature-inspired information flow structure as a basis for networked control.