Heterogeneity is the essence of the IoT paradigm. There is heterogeneity in communication and transport protocols, in network infrastructure, and even among the interacting devices themselves. Managing discovery of the different devices in such a paradigm is an extremely complex task. The typical solutions include an abstraction layer, commonly known as the middleware layer, that handles this complexity for the devices, thereby, allowing them to interact with one another. One major limitation of the existing middleware solutions is in their ability to allow for an easily configurable approach required to handle the tremendous scale of heterogeneous components in the IoT. The objective of this thesis is to develop such a highly configurable discovery middleware approach. The proposed approach aims to discover a variety of heterogeneous devices and services depending on a multi-level plugin layer, consisting of independent plugins that interact with each other based on the pipes and filters architectural pattern. To allow for the dynamic configuration of the middleware, a discovery configuration is developed. The output from the middleware includes a list of devices and their capabilities and is accessible via a web interface which can interact with a range of different clients. The proposed approach is validated on a scenario in a real-life environment.