This work focuses on proposing a method of effectively dealing with P2P-based service selection and composition, especially when handling a large number of peers along with their diverse qualities. The QoS-aware peer selection is one of the major challenges faced in order to guarantee the success and enhance performance of distributed computing. Since many peer candidates provide overlapping or identical functionalities, though with different QoS evaluations, selections need to be rapidly conducted to determine which peers are suitable to join in the requested composite service. The main contribution of this paper is proposing a P2P-based service selection model, in which peer's non-functional properties are modeled with Web service modelling ontology (WSMO), and where ant colony optimisation (ACO) technique is adopted to facilitate and enhance the QoS-aware peers' composition. We present experimental results to illustrate the effectiveness and feasibility of the proposed method.