HALF: Heterogeneity-Aware Lightweight Framework for Service Deployment

With the rapid development of the Internet of Things (IoT) and edge computing, embedded systems are facing increasingly complex service deployment demands in distributed applications. However, existing approaches such as containerization, script-based deployment, and static deployment suffer from poor adaptability and limited flexibility in resource-constrained and heterogeneous embedded environments. These limitations significantly hinder the adoption and promotion of microservice architectures in embedded systems. This paper proposes a Heterogeneity-Aware Lightweight Framework (HALF) for service deployment tailored to heterogeneous embedded environments. The framework features three key capabilities: (1) abstraction of hardware and software capabilities to enable heterogeneity awareness across nodes; (2) support for declarative service deployment constraints and constraint-based matching to improve deployment accuracy and automation; and (3) differentiated deployment strategies based on target node characteristics, supporting various deployment methods while unifying the external deployment interface. A prototype system has been implemented and validated on Linux, Reworks RTOS, and bare-metal STM32 platforms. Experimental results demonstrate that HALF can complete node registration and service deployment within a reasonable time, validating its practicality and applicability in heterogeneous embedded deployment scenarios.