Using Asynchronous Transfer Mode(ATM) which is a high-bandwidth, low-delay, cell switching and multiplexing technology, Broadband-Integrated Services Digital Network (B-ISDN) can support communication services of all kinds. To evaluate the performance of ATM networks, traffic source models to meet the requirements are demanded. We can obtain random traffic distribution for ATM networks by using the Cellular Automata (CA) which have effective random pattern generation capability. In this paper we propose an algorithm using 2-D LHCA to generate more effective random patterns with good random characteristics. And we show that the randomness by 2-D LHCA is better than that of the randomness by 1-D LHCA.