USC Sampling Based Approximate Aggregation System Disign And Implemention

In applications of wireless sensors networks(WSN for short), the aggregations ofdata, such as max, min, sum and average, are very important for users to get informationabout the monitored area. Till now people have proposed many aggregation algorithms.But those algorithms aim at the exact aggregation results and for that they collect toomuch data so that they cost too much energe. In practice, many applications only requireapproximate aggregation results so we could use approximate aggregation algorithms and"Samping based (?,δ)-Approximate Aggregation Algorithm in Sensor Networks"is one.Our major work is design and implement one approximate query system which baseson"Samping based (?,δ)-Approximate Aggregation Algorithm in Sensor Networks".Thisquery system get ? andδas inputs from users and return the approximate values to user.The whole system lie on TinyOS platform and include GPSR routing, USC sampling andgraphics interface. GPSR works for communication between sink mote and cluster headmote. USC sampling works for approximate sampling in clusters. Graphics interfacebases on Qt graphics frames library and it works on PC. System's workflow is like this:sink mote decide the sampling size and send it to cluster head motes via GPSR routing,then cluster head motes finish sampling via USC sampling and return the sampling resultto sink mote.In implemention of this system, we also implement GPSR,Trickle and CollectionTree Protocol. The main difficuty is in real WSN, design and implement good data struc-tureforthesethetheoreticalbase. WebuiltmanycomponentsbasedonTinyOS'sbottomcomponents. At the end of this paper, author gaves a show about system interface. Ex-periments show us the system can work well and it fulfils the system design need.