1 Abstract— In a pervasive computing scenario, a fundamental functionality is the discovery of available services: in fact, the first step in the long chain bringing an user to access a service is to find it. In the literature there are a number of service discovery protocols, each of them excelling in its particular field of application, and also a number of extensions to usual discovery architectures, especially to empower their descriptive capabilities. In any case it is missing an architecture able to solve the problem of the inter-working among different service discovery approaches – typically, each service discovery protocol has its own rules, and limitations – instead to have a common architecture able to use all actual protocols in the same way. In this paper we propose a service discovery framework which addresses this problem, introducing the concept of ontology – based multi-protocol service discovery: the proposed architecture, apart to use OWL to enhance, in the same common way, all descriptive capabilities of discovery protocols, is tailored to be compatible with all discovery protocols, due to the use of an opportune interface. This architecture has been deployed and tested using the SLP and UPnP service discovery protocols, integrated with the powerful descriptive functionalities offered by OWL. This work describes the results of the design activity which is being carried out within DAIDALOS II (Designing Advanced network Interfaces for the Delivery and Administration of Location independent, Optimised personal Services), a project granted in the European 6 th Framework Research Programme, within the IST (Information Society and Technology) thematic area. Index Terms— Service Discovery, Multiprotocol, Pervasive System, OWL, Semantic Service Description, Service Ontology. I. INTRODUCTION HE first phase to utilize an available service on a local network or in the Internet is to discover it. To address this issue many solutions have been developed: nowadays there is a plethora of discovery protocols and software architectures (SLP[4], UDDI [11], UPnP [2], Salutation [3], just to mention the most famous) that usually are the best in their specific ambit, but, at the same time, they don’t guarantee compatibility with each other. A service having a description model that allows a protocol to discover it, will not be discovered by nothing else but that specific protocol, because of the different description model, architecture or networks used by different protocols. This problem could be solved using the architecture developed by our research group within the DAIDALOS II project and presented in this paper. This architecture lays on a multi-protocol discovery architecture and use a shared semantic-based knowledge to describe services. The paper is organized as follows: section II explains how the entire multi-protocol service discovery framework works focusing on the main innovations; section III presents the state of the art of the most used service discovery protocols and introduces OWL as a semantic language for the service description; Section IV shows an implementation of the service discovery framework; section V concludes the paper with a pro and cons analysis of the proposed architecture. II. THE MULTI-PROTOCOL SERVICE DISCOVERY A lot of work has been carried out in literature on integrated solutions for service discovery where a new powerful protocol for service discovery has been invented (e.g. [16]), or a middleware platform has been design to be compatible with the existing discovery protocol (e.g. [17] and [18]). We followed a more feasible solution, which extends and comprehends the previously ones, based on the following principles: 1. to integrate service discovery protocols with a common language for service description, preserving the backward compatibility in the discovery phase using the formers and gaining a powerful service description from the latter. The use of an ontology-based approach to describe services permits to filter services in a much more accurate way. Thanks to the capability of a semantic language to describe relationships between services, the use of ontologies is also a cheaper solution in terms of memory allocation for the service description files (the redundancy decreases). 2. to extend the ontology-based architecture with a multi- protocol service discovery convergence layer. All the implementations of an ontology-based service discovery architecture are limited to only one protocol or self- standing. Adding the possibility to use different protocols to find services will enhance the recall rate of the service discovery architecture, because discovered services will be equal to the union of ones discovered by each protocol. An Ontology-Based Multi-Protocol Service Discovery Framework Silvano Mignanti 1 , Vincenzo Suraci 2 , Carmine Di Menna 3 University of Rome “Sapienza” - DIS Department - via Eudossiana, 18 - 00184 Rome (Italy) email: { 1 mignanti, 2 suraci}@dis.uniroma1.it; 3 carmine.dm@fastwebnet.it T