Statement of Purpose
Wireless sensor networks (WSN) have their irresistible elegance and attraction in structuring the Internet of Things. Abridging the invisible distance between human beings and incognizable environment, sensor networks provide accurate and valuable information of our surrounding environment. I would like to begin with my research plan to show my enthusiasm.

Research on Algorithm and Implementation of Topology and Clustering Protocol in WSN

Clustering is a basic technique for prolonging the system life and extending system coverage. Several protocols have been proposed, yet they still need improved and seldom of them have been put into practice. The implementation is based on the algorithm. Therefore, the process shall be divided into two parts.
1. To improve the clustering algorithm, several premises including energy spending in circuitry and transmission, network scale and frequency of tranceiving data should be broached in advance so that a mathematical model can be built. Previous clustering protocols such as LEACH, HEED, SPAN can be stimulated on this WSN model and their results can be analyzed. Then remodify the routines how they elect cluster heads and how they multi-hop from each other to the basestation with knowledge in Probability Theory and Graph Theory.
2. A huge gap lies between the emulation and practice. The protocol stack shall be divided into two layers: The MAC Layer mainly takes charge of basic reliable transmission job like frame synchronization, error control, anti-collision control, etc.; the Network Layer is responsible for packet switching and forwarding between cluster heads on the previous multi-hop routing scheme. Therefore it can adapts the structure of Zigbee which exploits IEEE 802.15.4 as controlling the MAC and PHY Layer. Two ways can be tried to apply its router strategy. First is to configure router tables on its Network Layer. It requires deep understading of the kernel of the protocol stack. Another is to carry out the routing strategy on the APP Layer with APS provided. Plus, control protocol might be needed in the same way how RTCP serves RTP in the transmission of stream media. In order to put all these into practice, significant efforts will also be spent on embedded operating system like Linux kernel.

I am determined to do theoretic and practical research in XX University

Interest and enthusiasm:
My freshman year was devoted to finding my true interest. I attended the “Dream-On” Academic Program where I learnt communication theories ahead of schedule and nonacademic pursuits like Java programming under Software Department. Neither of them intrigued my curiosity. However, Prof. Ling showed me the elegance of wireless sensor networks and it was not long before I became fully attracted by it. From then on, such enthusiasm has always been driving me to pursuing its further disciplines. I got a 3.37/4.0 GPA in electronics/computer/networks related courses, basing quite good foundations for my further research studies.

Rich lab experiences and solid programming background:
Following Prof. Ling accomplishing the project of reconstructing a whole network system of a high school with the concept of “Intelligent Campus”, I spent a lot of efforts on IoT systems and reviewing Computer Networks, a major course learnt previously. Out of the desire for releasing out a WSN application system, I joined the lab of Prof. Qian. The work has been described in my paper []. However, as I was quite inexperienced, a lot of practical difficulties had to be overcomed. Reading device manuals was even a molehill compared with follow-up programming and debugging procedures. Two entire days were spent in the lab because the serial communication was abnormal. Certain fields of the frame received were constantly 0xffff. Reduplicative single step debugging had to be carried out before I came to the final simple conclusion that the CHAR type defined of the field should be BYTE as the MSB contributed to its value. The joy of trouble shooting was impayable. Another trouble in this project occurred when I was writing the monitoring system software on PC where dynamic images were shaking severely, especially when dragging the window dialogs. Deeper inspection uncovered the reason that the workload of drawing method was quite massive that it required more execution time than usual everytime it repaints. Therefore I came up to the method of using double buffer following the strategy that the image was drawn first in the RAM and then copied to the screen, taking advantage that the RAM has a far quicker input/output speed. This double buffer strategy was applied to process the length-variable protocol frames retrospectively and also showed a good performance.

During the previous experience above, I realized that the key factor to a whole system is not on the upper application layer, but rather the lower ones where collision, poweroff or software failure takes place. So I joined Prof. Zhu’s group as one of the leaders of his research team working on the system of board sailing game environment monitoring and sports analysis, which, compared with the previous project, contained many more sensors and nodes, making it a good opportunity to perform deeper researches on lower layers. Therefore I took responsibility for the core gateway node development on ARM (Cortex M3) architecture in addition to the replay system software. Combining three types of communication: Zigbee, GPRS and 3G, I had to understand all frame formats of them. In order to control all the nodes sequentially, A polling strategy based on client/server mode was built which was synchronized by GPS modules obtaining benchmark time from satellites, avoiding the bottleneck effect which might occur if the network’s time benchmark was provided by a certain given node and improving the network’s reliability.

XXX School is my first choice:

It's a bit longer than I expected. Does it suit for a MS application?