The heterogeneous and resource-constrained nature of the Internet of Things (IoT) poses significant challenges to efficient and scalable service provisioning. Traditional service-oriented architectures often struggle to cope with these constraints, leading to high latency and increased failure rates. This paper introduces an Enhanced Context-aware Semantic Service Provisioning framework (E-CSSP) designed to address these challenges. E-CSSP leverages context awareness and semantic web technologies to enable dynamic service discovery, personalized service delivery, and optimized resource utilization within heterogeneous IoT environments. We present a practical implementation of E-CSSP using Constrained Application Protocol (CoAP) and HTTP protocols and evaluate its performance through simulation. The results demonstrate that E-CSSP achieves significantly lower response times (an average reduction of 68%, from 1.99 s to 0.64 s) and reduced failure rates (a decrease of 95.4%, from 22.0% to 1.0%) compared to traditional approaches. These findings highlight the potential of E-CSSP to enable efficient and scalable service provisioning in heterogeneous IoT environments.