WSNs typically comprise of sensing nodes with limited computational capability and onboard power. The sensed data in a WSN is transmitted from an individual node to a network sink in a multi-hop fashion. Since transmission costs are often several orders of magnitude larger than computational costs, the efficiency of the WSN can be improved by in-network data aggregation techniques. This is, however, problematic because the aggregation to be performed depends on the requirements of the end user/application, and is either unknown at the time of deployment or changes over time. Implementation of fixed aggregation algorithms limits the utility of the network. Software-based implementation of dynamic aggregation techniques offers the required flexibility but has significant processing overhead, especially when the size of the network increases. In this paper, we reduce the processing overhead by implementing dynamic data aggregation using reconfigurable cluster heads (RCHs) based on Field Programmable Gate Arrays (FPGAs). Such an implementation provides the necessary flexibility in data aggregation techniques demanded by real-time applications, while resulting in significant reduction in the query processing time and the overall power consumption in the network. The objective of the paper is to address the performance improvement in Wireless Sensor Networks (WSNs) through the use of reconfigurable cluster heads. Our results demonstrate that different data aggregation algorithms can be dynamically and efficiently implemented on the RCHs in run-time. The proposed approach is a crucial first-step towards the implementation of programmable WSNS.