The rapid growth of distributed real-time systems creates a need for cheap and available network solutions while still fulfilling the real-time requirements. In this paper we propose a method for less pessimistic delay analysis for packet switched first come first serve network, when knowing the intervals of possible message generation. Experiments show that the proposed method generates the expected results according to theoretical limitations of the experiment cases. The experiments also show that the proposed method could be practically used for non-trivial systems. Suggestions are given for future work on how to relax traffic requirements and how to cope with circular dependencies.