Overview

Cyber-Physical Systems (CPS) encompass systems in which the cyber world of computation and communication closely interacts with the physical world of sensors and actuators. These are highly networked and deeply embedded systems such as those found in modern day aircrafts, automotives, factories, medical devices, smart phones, electric grids, etc. From driver-less cars and air traffic management using sense and avoid, to plug-and-play operating rooms and smart electric grids that integrate traditional and renewable energy sources, intelligent automation of an enormous scale is finding its way into many of these systems.

Current interests of the CPS Research Group @SCSE,NTU are in the design of cyber-infrastructures for CPS, especially for systems with time-critical functionalities (i.e., real-time systems). The research themes can be broadly classified as follows:

  1. Real-Time Systems (scheduling algorithms and hardware designs)
  2. Model-based Design Methodologies (for resilience and operational efficiency)
  3. Energy Management (scheduling algorithms)

Ongoing Projects


Resilient Cyber-Infrastructure for Manufacturing

image-left Sensor proliferation and large-scale connectivity have enabled a variety of functionalities in CPS. However, connectivity also means that these systems operate in unreliable open environments, and hence resiliency to faults and attacks become important. This resiliency is fundamentally dependent on the resiliency of the cyber-infrastructure (communication network and computation nodes), which plays a central role of data delivery and execution of control. The objective of this project is to design a resilient cyber-infrastructure for such emerging CPS. A smart manufacturing testbed is also built to demonstrate the capabilities developed in the project.

Delta-NTU Cyber-Physical Systems Corporate Lab (National Research Foundation, Singapore and Delta Electronics Inc.)

Read More

Model-in-the-Loop Framework for Manufacturing

image-left With sensing technology becoming pervasive in manufacturing plants, large amounts of data are being generated in real-time. As a consequence, there is a need to effectively utilize this big data so that the desired objectives of Industry 4.0 such as predictive maintenance, root cause analysis and re-configurability, can be realized. The digital twin, obtained by using this big data together with plant and controller models, can be viewed as an accurate and time-synchronized characterization of the physical system in the cyber space. In this project, we aim to develop a tooling framework to realize the design and deployment of a digital twin in manufacturing with the following objectives: introducing a software architecture for the digital twin construction and application development, building a digital twin using heterogeneous modeling formalisms and online data integration, and providing a single source of truth for isolated applications via a well-formed interface.

Delta-NTU Cyber-Physical Systems Corporate Lab (National Research Foundation, Singapore and Delta Electronics Inc.)

Read More

Optimization Framework for Energy Management Systems

image-right In Singapore, the operation of buildings consumes about 37% of total electricity, making them one of the largest consumers of primary energy. Around 70% of the energy used in buildings is consumed by space cooling and dehumidification. Thus, the reduction and more efficient use of energy for cooling in buildings will provide a large leverage for climate change mitigation. District cooling systems that supply such cooling energy from centralized chillers via distribution networks within a geographical area have emerged as an efficient alternative to personalized air-conditioning systems. Although such cooling systems are in commercial use globally, significant operational inefficiencies and losses are often reported. Towards mitigating these inefficiencies, in this project, we develop a design methodology for formal modelling and operational optimization of district cooling systems.

Industrial PhD Grant (Economic Development Board, Singapore and Veolia City Modelling Center)

PhD Grant (ERI@N, NTU)

Read More

Distributed and Internet-of-Things (IoT) based Electricity Metering and Load Management

image-right The aim of this project is to implement the concept of decentralized electricity metering and load management using IoT capabilities in smart devices. The long-term objective is to remove household- or company-level smart electricity metering and replace it with aggregated reporting of all devices in the household/company environment. Moreover, this approach will also enable these electricity-consuming devices to participate in a deregulated electricity market. By developing approaches from the blockchain technology domain towards application in embedded devices, a trustless authentication mechanism for provisioning and verifying the compliance of these devices will be made possible. Our investigation of how to achieve a fully decentralized energy grid with smart contracts, binding all grid participants to system-level goals, will contribute to bringing smart grid functionality to a level of granularity previously not possible.

Intra-CREATE Seed Grant (National Research Foundation, Singapore and TU Munich)

Read More

Assured-Safety Architecture for Machine Learning based CPS

image-right Machine learning (ML) techniques are increasingly applied to decision-making and control problems in Cyber-Physical Systems among which many are safety-critical, e.g., chemical plants, robotics, autonomous vehicles. Despite the significant benefits brought by ML techniques, there are various factors that can impede the achievement of ML safety. For example, 1) expressive ML models such as deep neural networks (DNN) are typically considered to be non-transparent, behaving as a “black-box” and lacking interpretable knowledge representation; 2) The empirical risk minimization approach used to train ML models reduces the probability of false prediction on the assumption that the training samples are drawn from the actual underlying probability distribution of the population; 3) Formal verification requires a specification of the property of interest, i.e., a precise, mathematical statement of what the system is supposed to or not supposed to do. However, it is difficult to come up with such a formal specification for ML-CPS. In this project, we focus on the design of novel techniques to improve the safety of ML-CPS.

Grant (ERI@N and Cyber-Physical Systems Corporate Lab, NTU)

Read More


Past Projects


Mixed-Criticality Scheduling Algorithms

image-right With increasing functionality and automation in real-time systems, the computational demand is steadily increasing, while resources available for servicing this demand are limited due to size, weight and power restrictions. As a result a module integration trend has emerged, so that many different applications are being cohosted on the same processing unit. Some of these applications are extremely critical for the correct behavior of the system (e.g., collision avoidance in automotive), while some others are relatively less critical (e.g., hill assist in automotive), thus giving rise to mixed-criticality real-time systems. In this project, we focus on the design of single-core as well as multi-core scheduling algorithms for such systems. We also develop practical workload models to characterize such systems, and evalute the same using an automotive testbed.

Tier-2 Grant (Ministry of Education, Singapore)

Read More

Cache Designs for Timing-Predictability

image-right Cache memory plays a key role for performance improvement and energy efficiency in modern computer systems. However, when used as a shared resource in a multicore system, cache memory becomes the key source for timing unpredictability. As predictability is a major concern in real-time systems, this project designs cache memories to address this issue. We focus on the development of hardware techniques, specifically cache replacement policy designs, to improve the predictability of both private as well as shared caches on multicores.

Tier-1 Grant (Ministry of Education, Singapore)

Read More