Acompressionalgorithmisproposedinthispaperforreducingthesizeofsensordata.Byusingadictionary-basedlosslesscompressionalgorithm,sensordatacanbecompressedefficientlyandinterpretedwithoutdecompressing.Thecorrelationbetweenredundancyofsensordataandcompressionratioisexplored.Further,aparallelcompressionalgorithmbasedonMapReduce[1]isproposed.Meanwhile,datapartitionerwhichplaysanimportantroleinperformanceofMapReduceapplicationisdiscussedalongwithperformanceevaluationcriteriaproposedinthispaper.Experimentsdemonstratethatrandomsamplerissuitableforhighlyredundantsensordataandtheproposedcompressionalgorithmscancompressthosehighlyredundantsensordataefficiently.
