A hierarchical high-throughput and low power architecture for longest prefix matching for packet forwarding

Internet routing must employ a network processor with very fast routing mechanism to maintain required high throughputs. Each packet must be forwarded to its most suitable hop based on the information stored in the routing table. Therefore to select the best suited prefix (highest subnet mask) amongst all the matched ones, it is required for the IP networking to perform Longest Prefix Match. To meet the requirement of high-speed (multi-gigabits/sec) IP processing, a low-power high-throughput lookup architecture has been presented in this paper to reduce power consumption and to increase the efficiency of finding the destination address at a faster rate. This architecture avoids the necessity of arranging prefixes according to their prefix lengths as in existing ternary CAM solutions and also provides simple and fast updates. Additionally, the proposed architecture reduces the TCAM area requirements by involving both BiCAM and TCAM for storing prefixes and therefore saves power. It further saves power through smart search operations by decomposing entire table into sub tables, where only one sub table is active at a time.