Millimeter wave (mmWave) technology can meet the requirements of large amount of data communications among intelligent devices for indoor scenario. With high requirement of throughput, reducing interference in an enclosed region is still a challenge. For indoor scenario, the human bodies are considered as the main blockages besides conventional blockages. In this paper, we present a new system model by exploiting the self-blockage model to capture the effects of human body, while using multi-ball Line of Sight (LOS) link state model to describe the conventional blockage. The combination of the two components provides a more comprehensive and accurate expression of indoor obstruction. We give a closed expression of coverage rate to analyze the system performance. Simulation results show that the proposed model is highly accurate to describe the distribution of blockages for indoor case. Then, we formulate the resource allocation as an optimization problem under this scenario. The target of this optimization is to achieve the maximum throughput by minimizing interference. To reach this goal, an improved Powell Multi Vertex Coloring algorithm (PMVC) is proposed. The throughput raises apparently when we modify this scheme to put more flows in different time-slots evenly. Numerical experiments show that the improved algorithm can achieve higher system throughput than traditional greedy algorithm and Powell Vertex Coloring (PVC) scheme.