With  the development of the functional diversity and intelligence in automotive systems, the number of ECU increases drastically, so it brings enormous pressure to cost reduction. Based on the TTP bus, ECUs construct a distributed architecture  characterized by periodic task activation, TDMAbased message sending and preemptive FPS. Realtimeness and cost reduction are two key problems of this architecture, especially for the increase of complexity, which poses great challenges to resolving these two problems. In this paper, we propose an optimized approach which puts geometric programming and schedulability analysis together, in order to simultaneously finish the period assignment and bus access configuration, meet endtoend deadline constraints and minimize CPU utilization. Our approach is applied in an industriallevel case study and the simulation results show it is effective and scalable.