Towards Ameliorating the Problem of Packet Dropping in IDS using P System Model on GPU
​
Rufai Kazeem Idowu, Ravie Chandren M., and Zulaiha Ali Othman
​
​
Abstract. As the number of internet users grows exponentially everyday, the attacks and intrusions experienced on different networks equally multiply rapidly! Most of the current Intrusion Detection Systems (IDSs) are confronted with the challenge of coping with this high volume of traffics because they have low processing throughput. The overall consequence of this short-coming is what is known as packet dropping. P system, which is otherwise called membrane system, is a maximally parallel, non deterministic and highly distributed model inspired by the functioning of living cells in Biology. Therefore, in exploring the parallelism advantage of both membrane system and Graphics Processing Unit (GPU), the paper presents an attack detection P system implemented on GPU to minimally decrease the negative impact of packet dropping which usually emanates from busy networks. On evaluation with KDD Cup dataset, our model achieves high throughput average of about 50000p/s and classification accuracy of 95.8%.