In this paper we analyzed Chandy-Misra null message algorithm and proposed and implemented autonomous grouping technique to improve the performance. Our primary goal is to minimize the synchronization overhead in conservative parallel simulation by reducing the number of null messages over the network. Exchange of large number of null messages effect the performance of simulation and liable to congestion. Different techniques has been proposed most of them are simulation dependent. The performance of null message algorithm highly depends upon lookahead value. The smaller the lookahead the more number of messages transmitted during the simulation. We presented an autonomous grouping approach to reduce the number of synchronization messages sent out of the system.