Time Division Multiple Access (TDMA) protocols are considered an energy-efficient solution to prolong wireless sensor network lifetime. However, their drawbacks such as the complexity of slot assignment and schedule maintenance, clock synchronization and low channel utilization during low traffic conditions are to be overcome in a good and efficient way. In this paper we present On-demand Convergecast Sensor MACS (OCSMACS), a centralized and adaptive multihop scheduling-based TDMA protocol for Wireless Sensor Networks. OCSMACS achieves high energy efficiency by adopting a novel requests aggregation mechanism for adaptive slot assignments such that all requests generated by active nodes are aggregated in a very few small size packets. In addition, knowledge of data correlation is integrated in the design of underlying multihop scheduling schemes to improve energy efficiency and extend network lifetime. Furthermore, OCSMACS adopts an energy-efficient protocol called PROGRESSIVE for the topology discovery and setup phases. The performance of OCSMACS is compared against existing protocols based on simulations using ns-2 and the results show that OCSMACS outperforms these protocols in terms of energy efficiency and network lifetime. Also, the integration of correlation knowledge in the scheduling process improves performance and extends network lifetime.