| Abstract: |
CONCLUSIONS AND FUTURE WORK7.1. IntroductionMost security researches focus on the field of information assurance, which protects data using techniques such as authentication and encryption. However, information assurance assumes that the devices responsible for encryption, forwarding, and sending packets are trustworthy. So, we focused on network infrastructure (e.g. routers, servers) security because they are compromised by malicious adversaries.Routing table threats constitute a challenging problem since routing table forms the basis of the Internet and any corruption of it may lead to dangerous consequence. Secure Link State Protocol (SLIP) is intuitively attractive. It is used to detect and locate the malicious nodes in a network. Unfortunately, the Internet is not reliable, and SLIP makes several implicit assumptions that may not hold. This thesis explores SLIP, its deficiencies, and how it can be improved. The primary benefit is a computer simulation of the SLIP to get empirical results in the form of performance data and proof of concept evidence of both SLIP’s advantages and its shortage in case of inactive attacks and some cases of proactive routing attacks. This helps us to develop a new scheme to overcome the shortage of SLIP.7.2. ConclusionThe present work has developed models for the mechanisms involved in the SLIP and subjected this protocol to computer simulation, through which it is shown that the major disadvantage of the SLIP is its failure in protecting the network in the case of inactive routing table attacks and in the cases of proactive attacks in which different information about a link cost is sent by different nodes. The present work provides a new scheme ”Modified Secure Link State Protocol (MSLIP)” that takes the advantages of SLIP and adds the necessary operations to protect the internet infrastructure in all the cases of inactive and proactive routing attacks. The added operations involve obtaining an estimate of the link cost by inquiring a number of selected nodes about the link cost. Each of these nodes, in turn, carries out cost measurement and then sends a reply about the link cost, which is considered as a vote. Taking the average of the link costs provided by the (voting) nodes enables to correctly make a decision about the malicious node. The MSLIP is modeled and subjected to computer simulation through which is shown that it is able to correctly determine the malicious node in both cases of inactive and proactive routing table attacks.Under non-malicious environment the variation of fault detection time is studied with respect to those of graph density and fault rate. It is found that, fault detection times of both MSLIP and SLIP are the same, and the fault detection time under both MSLIP and SLIP is significantly more than that of link state when the graph density is low (Under malicious environment, the variation of attack detection probability (?) is studied with both number of malicious nodes in the network and the node degree. In both cases of SLIP and MSLIP,? decreases with increasing number of malicious nodes in the network. However, MSLIP works better than SLIP with increasing the number of malicious nodes. The probability to detect an attack is greater in case of MSLIP than in SLIP with the increase of node degree.7.3. Future WorkThere are some suggestions that to be done in the future in order to make the MSLIP more general and efficient:? Since the MSLIP is built on the SLIP, it is considered to be scalable as the SLIP. So, the scalability of the MSLIP is to be measured by using more than subnet to prove its efficiency.? The protocol takes a long period of time to perform synchronization algorithm that is carried out using the principle of voting. The synchronization algorithm has a worst-case running time of O (n3). Because of its high running time, it is valuable to try minimize this period.
|
|
|