Enhancing IoT Service Discovery through Semantic Name-based Forwarding

Named Data Networking (NDN) has emerged as a promising communication paradigm for the Internet of Things (IoT). In NDN, application-level service names, expressed as Uniform Resource Identifiers (URIs), are used directly at the network layer to facilitate service discovery and retrieval. Forwarding decisions at each NDN router rely on the longest-prefix match between the name carried in the incoming service request and the entries in its Forwarding Information Base (FIB). However, in highly dynamic and heterogeneous IoT environments, client applications, interested in accessing a given IoT service, may not always know the exact service names in advance. Additionally, different service providers, each using distinct namespaces but offering functionally equivalent services, may be capable of fulfilling the same request. This paper explores an alternative name-based forwarding approach that leverages semantic similarity to improve the chances of successfully resolving incoming IoT service requests. When an NDN node cannot find an exact name match, it applies a semantic similarity-aware mechanism to identify the most relevant forwarding path in the FIB toward a potential IoT service provider. This approach is implemented using deep learning models for sentence embedding, seamlessly integrated into the NDN forwarding fabric. Evaluation results demonstrate that semantic-aware forwarding significantly enhances service discovery in scenarios where exact name matching fails, while maintaining low processing times.