Grid computing aggregates distributed computing resources to execute computationally complex jobs. The selection of resources in a Grid system involves finding and locating resources based on users' requirements. Identifying an appropriate resource selection mechanism for Grid jobs is a major concern because a suitable mechanism helps to schedule and allocate resources better. The resource discovery techniques proposed over the years can be categorized into three major types-centralized, hierarchical and decentralized. The performance of each type can be measured in terms of scalability, load balancing and fault tolerance. The resource discovery based on centralized and hierarchical techniques is recommended for small and medium size Grids. Due to limitations such as poor scalability and single point of failure in centralized and hierarchical approaches, a decentralized resource discovery is usually preferred for large size Grids. Lack of coordination between users and providers in a highly decentralized heterogeneous Grid environment often results in user jobs failing in finding relevant resources. One of the reasons for rejection of jobs is the usage of fixed schema between users request and providers availability which causes high possibility of missing relevant resources. We present a semantic decentralized resource discovery model by using P2P Chord protocol to improve job success probability and to enhance utilization of resources. Preliminary simulations carried out using GridSim and PlanetSim simulators show improved success probability for complex jobs and also show promising results with enhanced utilization of resources.