Orchestration of Interdependent Tasks in Edge Computing Systems

Abstract

The rapid expansion of Internet of Things (IoT) devices has redefined computing demands, necessitating low-latency, real-time, and bandwidth-efficient solutions. Although cloud com- puting offers scalability and processing power, its centralized nature introduces latency and bottlenecks—issues especially critical in applications like autonomous systems, smart health- care, and industrial IoT. Edge computing addresses these limitations by relocating computation closer to data sources, reducing transmission delays and supporting real-time processing. However, the shift from monolithic to microservice-based applications introduces new challenges for or- chestrating workloads across distributed, heterogeneous edge environments. Existing orchestration methods are often static or tailored for centralized systems, lack- ing adaptability to mobile contexts, dynamic resource availability, and network fluctuations. This work proposes a heuristic-driven dynamic partitioning algorithm designed to orchestrate microservice-based workloads intelligently across edge nodes. It considers real-time factors such as node proximity, energy constraints, and task dependencies to optimize latency, re- source utilization, and responsiveness. The approach is evaluated using PureEdgeSim, a simulation toolkit supporting mobility, energy-awareness, and microservice orchestration. Results show a clear reduction in end-to- end latency, improved load balancing, and better adaptability under dynamic conditions. These outcomes demonstrate the promise of heuristic-based orchestration in enabling ef- ficient, scalable, and context-aware edge computing.