Cyber-physical systems serve as the foundation of modern Industry 4.0 and 5.0 manufacturing, wherein Service-oriented architectures (SOAs) and System of Systems (SoS) are widely established concepts. Entities within such systems often comprise machines and software components, characterized by autonomous actions while providing and consuming services to fulfill their designated tasks. With the growing demand for mass customization, consumer entities require more comprehensive information about the requested services to maintain effectiveness and efficiency. To achieve this, a more detailed Quality of Service (QoS)-based service selection becomes imperative. This paper introduces a reference architecture for the implementation of QoS-based service selection in SOA-based SoS frameworks as an integral part of the service orchestration process. The proposed enhancements to the QoS management system address the evolving requirements arising from contemporary Machine-to-Machine (M2M) environments characterized by high levels of autonomy.