This paper describes an optimal traffic data archive scheme where the maximum information of the original data can be preserved using less storage space. Compared with traditional equal-width methods where compact data isobtained by aggregating source data at fixed intervals, the optimal scheme uses varying intervals to aggregate data at different levels based on the variations within the source data. The resultant scheme is optimal in terms of information conservation, i.e. the errors between the source data and the optimal compact data is the smallest. Operational traffic data havebeen used to test two proposed optimisation schemes, local and global schemes. It was found that under the local optimisation scheme, the size of archived data can be reduced 6-fold compared with traditional equal-width schemes. The size of the archived data can be reduced by 3 times if the global optimisation scheme is employed. For the covering abstract see ITRD E134653.
Abstract