| Viruses have become a major challenge to security and connectivity. Computer viruses travel through networks looking for any amount of vulnerability to compromise systems. As society's reliance on the Internet increases a more robust antivirus solution is required. In this thesis, a novel approach is presented to detect viruses on a network. The approach uses a fast searchable memory to scan data for viruses in the protocol stack at line speed. In particular this thesis has provided an architectural framework by examining two related cases. One case focuses on maximum speed and lowering power consumption while the other focuses on both lowering real estate usage and power consumption. For both of these cases, line speed scanning has been achieved. Detection accuracy and throughput measurements have been obtained by simulation using Icarus Verilog, while power and real estate utilization were obtained using Spectre simulator by Cadence. The results show that the hardware solutions presented can provide better network security than both a software solution and previously developed hardware solutions. The throughput was calculated at 2 Gbps, however, simulation results have shown rates upwards of 3.5 Gbps, which is comparable to current hardware antivirus solutions. This comes at a vast reduction in power dissipation: up to 67% for case I, 77% for case II and real estate reduction of up to 40% in case II. |
