The overall performance of an intrusion protection system depends not only on the packet header classification and pattern matching, but also on the post-operative determination of correlative patterns of matched rules. An increasing number of patterns associated with a rule heighten the importance of correlative pattern matching. This work proposes a TCAM-based smart architecture that supports both deep pattern-matching and correlative pattern-matching. The proposed architecture overcomes the difficulties in implementing TCAM when the patterns are very deep and the rules for packet payload involve many patterns whose positions lie within a range. A real case payload is simulated using a Snort 2.3 rule set and simulation results demonstrate the feasibility of the proposed architecture in supporting a high-speed and robust intrusion detection and prevention system.