The growth of the Internet as a large scale media provisioning platform has been a great successstory of the 21st century. However, multimedia applications, with their specific traffic characteristicsand novel service requirements, pose an interesting challenge in terms of discovery,mobility and management. Furthermore, the recent impetus to Internet of things has made it verynecessary, to revitalize research in order to integrate heterogeneous information sources acrossnetworks. Towards this objective, the contributions in this thesis, try to find a balance betweenheterogeneity and interoperability, to discovery and integrate heterogeneous information sourcesin the context of Future Internet.Discovering information sources across networks need a through understanding of how theinformation is structured and what specific methods they follow to communicate. This processhas been regulated with the help of discovery protocols. However, protocols rely on differenttechniques and are designed taking the underlying network infrastructure into account. Thus,limiting the capability of some protocols to cross network boundary. To address this issue, thefirst contribution in this thesis tries to find a balanced solution to enable discovery protocols tointeroperate with each other as well as provide the necessary means to cross network boundaries.Towards this objective, we propose ZigZag, a middleware to reuse and extend current discoveryprotocols, designed for local networks, to discover available services in the large. Our approachis based on protocol translation to enable service discovery irrespectively of their underlyingdiscovery protocol. Although, our approach provides a step forward towards interoperability inthe large. We needed to make sure that discovery messages do not create a bottleneck for thenetwork.In large scale consumer oriented network, service discovery messages could render the networkunusable. To counter this, ZigZag uses the concept of aggregation during the discoveryprocess. Using aggregation ZigZag is able to integrate several replies from different servicesources supporting different discovery protocols. However, to customize the aggregation processto suit once needs, requires a through understanding of ZigZag fundamentals. To this end, wepropose our second contribution, a flexible policy language that can help define policies in aclean and effective way. In addition, the policy language has some added advantages in terms ofdynamic management. It provides features like delegation, runtime time policy management andlogging. We tested our approach with the help of simulations, the results showed that ZigZag canboth reduce the number of messages that flow through the network, and provide value sensitiveinformation to the requesting entity.Although, ZigZag is designed to discover media services in the large. It can very well be usedin other domains like home automation and smart spaces. While, the flexible pluggable modulardesign of the policy language enables it to be used in other applications like for instance, e-mail.
