In this paper, our aim is to obtain optimum size of data packet for wireless communication. Because of limited energy of each source, we try to reduce an energy consumption by packet size optimization. There are several issues in variable packet size such as hardware limitations, additional overhead and resource management cost issue, so we use fixed packet size optimization. In the context of energy constrained sensor nodes, the energy consumption of starting transients can be quite large. Energy consumption in speed-up energy and start-up energy are more than energy consumption of the data aggregation. If the number of start-up increases then total energy required is more for sent data from sender to receiver. Hence, packet size affects energy consumption. Using binary BCH codes, it is possible to significantly improve the energy efficiency. When ARQ and FEC are considered, for higher values of signal to noise ratio (SNR) threshold, ARQ apparently exceeds the FEC codes but for lower SNR threshold values, the energy consumption can be decreased using FEC codes. However, longer packet sizes are less energy consumption per useful bit upon adequate link quality is guaranteed. But when an insufficient link quality then we have to choose small packet size due to increase reliability. In this paper, we have used routing based on number of affecting parameters like signal to noise ratio, distance between sources and sink along with left over energy level of nodes. We considered the concept of left over energy level of nodes that results in a step towards the increase lifetime of network. Finally, our goal is to optimize packet length with decreased energy consumption per bit and increased network lifetime. Furthermore, an optimization of packet is determined based on objective function, i.e., packet throughput and energy function.