TY  - JOUR
T1  - Fuzzy Based False Feedback Mitigation in Broker Architecture Using Linear Interpolation
AU - , P. Varalakshmi AU - , S. Thamarai Selvi AU - , Sarah Wazirah AU - , S. Ramakrishnan AU - , S. David Paul Rufus 
JO  - International Journal of Soft Computing
VL  - 2
IS  - 4
SP  - 513
EP  - 519
PY  - 2007
DA  - 2001/08/19
SN  - 1816-9503
DO  - ijscomp.2007.513.519
UR  - https://makhillpublications.co/view-article.php?doi=ijscomp.2007.513.519
KW  - Trust-index
KW  -feedback
KW  -interpolation
KW  -fuzzy logic
KW  -overhead
AB  - In a grid environment where remote services/resources are accessed, the trust-index of an entity is of utmost importance. In the broker architecture of the grid environment, the grid is divided (logically) into multiple domains where in each domain; multiple entities i.e., Consumers and Resource Providers (RP) co exist along with a number of brokers. The broker acts as a middleman for the transactions committed between the entities. Each entity will be associated with more than one broker to ensure reliability and eliminate single point of failure. The brokers’ task is to select an appropriate Resource Provider to provide a resource to the consumer’s request. After the transaction is committed, the consumer will submit a feedback to the broker. This feedback will be evaluated by the broker for genuineness. By doing so, the trust-index of the RP is safe guarded against dishonest ratings by the consumer. If the feedback of a particular transaction is found to be dishonest, the consumer’s trust index is reduced. In order to predict a rating to replace the dishonest feedback, we propose one-dimensional linear interpolation. This pseudo-feedback can be used to compute the new trust-index of the RP. We have used Fuzzy Logic to compare the dynamically varying trust-indices of both the RP and consumer. In our fuzzy model, the variations in the trust-indices of these entities are moderate than that of those obtained from our probability model when the consumer acts maliciously. Also, we have proposed an efficient method to effectively reduce message overhead that results when the feedback of every transaction is verified for genuineness.
ER  -