AnyComputing: A Novel Architecture for Orchestrating Various Computing Paradigms for Adaptive Workload Management

The proliferation of physical objects and intelligent systems, along with their inherent differences, such as levels of device mobility, generated data volume, required storage memory, and complexity in data processing, presents a significant challenge in achieving real-time and optimal responsiveness to the Massive volume of sent requests from these systems. Depending on the nature of each request and its requirements, effective processing must be by available and suitable computational platforms. This situation necessitates the urgent development of novel ideas and frameworks in request orchestration and processing by suitable computational platforms to enhance system performance and efficiently utilize available computational resources. In this article, we propose an integrated computational architecture spanning from edge devices of users and intelligent things to cloud servers. This design encompasses the reception, processing, and response to various computational and storage requests, considering specific requirements, through computational units deployed within this span. Our objective in devising this architecture is to establish mechanisms for categorizing diverse computational requests from users and things, effectively allocating resources to them, thereby reducing response time, optimizing bandwidth utilization and other available resources, managing latency for real-time requests, and minimizing energy consumption in user devices and intelligent objects.