The amount of different networks and services available to users today are increasing. This introduces a need for ways to locate and sort out irrelevant services in the process of discovering available services for a user. In this paper, we describe and evaluate a prototype of an automated discovery and selection system, which locates services that are relevant to a user based on his/her context and the context of the available services. The prototype is based on a multi-level, hierarchical system approach and introduces entities called User-nodes, Super-nodes, and Root-nodes. These entities separates the network into domains that handle the complex distributed service discovery, which is based on dynamically changing context information. In the prototype, a method for performing context-sensitive service discovery has been achieved. The service discovery part utilizes UPnP, which has been extended in order to increase network scalability. The experimental analysis of service discovery times, based on different scenarios is used to optimize parameter settings of the service discovery system in order to achieve short response times.