In this paper, we proposed a CGR-QV algorithm in order to solve the problems of low message delivery rate and low through-put problems in deep space communication networks, which are caused by the high delay, asymmetric data transmission rate, and fre-quent topology switching characteristics. The proposed method separate different message types in the queues, establish the relationship between TTL, wait time and Hurst to prevent the phenomenon of starvation and redun-dancy. In view of the impact of congestion on the network, improved algorithm use the Elastic load balancing strategy establish the relationship between optimal thresholds α and network performance reliability. Simulation result show that the algorithm has improved in throughput as about 19％, message delivery rate improved as about 7.5％ and bandwidth rejection rate decreased as about 4.7％, the algorithm can balance the impact of topology updates and resource consumption better in space communication networks.