In Wireless Sensor Networks (WSN) sensor nodes are deployed in a region to sense the information. These sensor nodes sense the similar information and sends it to sink node. This thing leads to redundancy at sink node. Sink node wastes most of its energy in processing redundant packets. To save the energy of node in order to prolong the network lifetime there is need to eliminate redundancy. Data aggregation is a process in which intermediate node receives multiple input packets performs aggregation and produce single output packet in the network. This process will reduce the number of redundant packets in the network. But redundancy sustains reliability. Therefore there is need to maintain redundancy but it should be up to an adequate level. In this paper we have focused on different issues in data aggregation process such as delay, redundancy elimination, accuracy and traffic load and mentioned various methods to solve those issues and then we compared some data aggregation techniques based on strategy, delay, redundancy, average energy consumption and traffic load. Further we have proposed a model based on our study which performs data aggregation at multiple levels and not only maintains the tradeoff between energy conservation and reliability but also addresses all the issues in data aggregation technique. © 2015 The Authors. Published by Elsevier B.V.