Progressive Transmission Of Massive Vector Data Over Networks

Presently the rapid increase of vector data and the bandwidth limit of the networkhas greatly restricted high-precision data transmission and access and stunted thefurther development of WebGIS. So how to access quickly to large scale vector dataunder the limited network bandwidth has already been a focus of study. Focusing onthe above problems, this paper has made a study of network progressive transmissionmethods of the large scale vector data. In this paper we used some key techniques,through which we have achieved some goals that are displaying while transmittingdata, maintaining consistency in topology, reducing users’ waiting time and improvinggreatly the efficiency of data transmission. And users can stop transmission at anymoment during walking in the three dimensional scene, which helps to avoid uselessdata transmission to save time.In this paper we firstly analyzed the problems, solutions and principles of networkprogressive transmission of vector data, then stated the research status in China andabroad. And then we analyzed comprehensively two common methods of integratedcurve algorithm: Douglas-Peuker algorithm and Visvalingam-Whyatt algorithm.Based on M-algorithm and Voronoi map, we declared our simplified algorithm in thispaper, which keeps topological consistency in the simplification process and avoidsintersect and self-intersect. During the simplification process the new bornhierarchical data makes the transmission from coarse data to fine data, which meetsthe users’ needs of access to data. Subsequently we introduced several key techniquesin progressive transmission of vector data: Multi-threaded concurrent tasks,prefetching technique,LOD management and buy out of memory. Multi-threadedconcurrent tasks refers to parallel processing of single task (visibility computation,draw and disk operation). We can speed up the parallel processing usingsingle-processor or multi-processor multi-threading technology. During the roamingprocess we can equalize the disk operation using prefetching technique because ofnon-balanced distribution. We use LOD technique to calculate the contribution of one map to the last image and then decide the precision of the map. This method caneffectively control the complexity of each frame of geometric data. When the amountof memory consumed reaches to critical value, it needs to select the temporarilyunused data and throws it out from memory. Combining the above four techniqueswith out-of-core technique, we completed the large scale network progressivetransmission of vector data. At last we stated the TCP/IP and Socket technique andtransmission scheme, format and solution. The experiments show the stated methodsnot only are feasible, but also effective, which solved the problems of large scalevector data transmission, reduced the users’ waiting time, avoided data redundancyand improved the transmission efficiency.