Model-driven Middleware Specialization Techniques for Software Product-line Architectures in Distributed Real-time and Embedded Systems

Product-line architectures (PLA)s are an emerging paradigm for developing software families for distributed real-time and embedded (DRE) systems by customizing reusable artifacts, rather than handcrafting software from scratch. To reduce the effort of developing software PLAs and product variants for DRE systems, it is common to leverage general-purpose ñ ideally standard ñ middleware platforms whose reusable services and mechanisms support a range of application quality of service (QoS) require- ments, such as predictability and low end-to-end latency. While standard middleware provides generality and e xi- bility to support many types of PLAs and product variants, standard middleware implementations often incur unneces- sary footprint overhead and lack optimizations needed to meet the QoS needs of PLAs and product variants for DRE systems. This paper describes systematic model driven de- velopment (MDD) techniques for specializing implemen- tations of standards-based, general-purpose middleware to support the application-specic QoS needs of different product variants created atop a PLA. Our preliminary re- sults show that implementations of standard middleware can be specialized transparently to better meet the QoS needs of PLAs and product variants, without compromising standards compliance.