In content-based publish/subscribe systems it is difficult to satisfy end-to-end delays for each subscriber in the presence of bandwidth constraints. To tackle this problem, we arrange the subscribers in a tree structure such that the individual latency requirements are fulfilled and for scalability divide the publishers into clusters. Here we address the clustering of publishers and to organize the clusters, we look at the advertisement and the degree constraint of each publisher. In this thesis we developed different clustering strategies, analysed them and verified our results experimentally.