Visible to the public International Workshop on Trustworthy & Real-time Edge Computing for Cyber-Physical Systems (TREC4CPS 2019)

Visible to the public 

  A B O U T   T R E C 4 C P S  

The increasing proliferation of Internet of Things (IoT) is giving rise to an ever-increasing volume of data being generated by the IoT sensors that reside at the edge of the networks. Of specific interest to us are CPS applications found in a wide range of societal and environmental applications where the streamed information is fed to decision makers, and where time is crucial for the quality of the decision. For instance, emergency-response and smart-transportation systems are prime examples of multi-domain smart and connected community applications residing at the edge of networks. Conventionally, such systems have been implemented using centralized architectures. However, as the scale and community penetration of these data driven applications are growing, the challenges of centralized architectures become apparent; for example, lack of scalability, existence of single points of failure, and saturation of communication resources. Moreover, the sporadic and uncertain arrival patterns for the IoT data streams challenges real-time stream processing because resources must be provisioned on-demand to fuse multiple temporally-unsynchronized data streams. Moreover, the temporally sensitive nature of the data, the limited resource availability at IoT devices, and the large volumes of generated information make it problematic to always stream these information to centralized cloud data centers, which may be multiple network hops away from the IoT devices and connected via links with fluctuating bandwidth and jitter, leading to variable delays that are detrimental to the cyber physical systems.

To address these issues, the community has been moving towards edge computing solutions that promise to enable city-scale, extensible smart systems that make best use of available information, network resources, and computing resources, including cloud computing resources. In edge-centric deployments, effective app and system management is critical due to the need to add/remove resources seamlessly, handle failures gracefully, and upgrade/reconfigure distributed applications. Furthermore, since the distributed system is made up of a collection of IoT devices and other hardware, a single-device centric solution is not amenable to resolve the challenges. Rather, an aggregate computing approach is needed. Within this context, addressing the computing challenge requires time-bounded elastic and on-demand, distributed multi-resource provisioning, application configuration and failure management. Existing techniques for resource management of edge computing, however, tend to focus on provisioning only one type of resource at a time and seldom consider the problem holistically.

Another challenge with edge computing solutions is the problem of the trustworthiness of analysis results. Lack of effective assurance mechanisms have brought us to the current reality, which includes susceptibility to data-integrity and information-leakage attacks as shown by a large number of recent incidents even in the centralized information flows. The problem is expected to be much worse in the near future due to tight resource constraints (which prevents deploying standard security solutions) and to the widespread diffusion of IoT (which multiplies the chances of attacks). Effective strategies will have to study trade-offs between security, privacy, trust levels, resources, and performance. Additionally, we urgently require comprehensive exemplar applications and data cases that show how these problems are being studied by industry and the research community.