Abstract: In the previous work， we have constructed the Optimal Multiple Threshold Channel Reservation (OMTCR) scheme to configure the bandwidth resource in multiservice LEO satellite networks from connection level.It has been verified that the OMTCR can efficiently trade off the QoS of users and the system revenue. In this paper， we solve the optimal threshold vector of the OMTCR policy based on Genetic Algorithm (GA). The basic GA has been repaired to adapt to the particular feasible points and the objective function of the OMTCR problem. To cope with more service type and larger bandwidth， we developed an iterative GA optimizing policy with small initial population. Extensive simulation results have shown that， the modified GA significantly reduced computation time while achieving comparable performance of exhaustive search in both QoS unconstrained and QoS constrained system revenue loss optimization model. The sufficient accuracy and reasonable runtime of the modified GA ensured the practicality of OMTCR scheme.