我想在一个车辆的驱动模式创建的签名。我已收集到的读数和绘制数据在3D，通过考虑时间坐标中的X轴，而纬度和经度在其他两个轴。这是非常重要的，我需要遵循时间轴，并采取了其他两个数字和COM preSS它，就像散列。任何建议或想法，我在哪里可以开始了？

I would like to create the signature in the driving pattern of a vehicle. I have collected the readings and plotted the data in 3D, by considering "time" co-ordinate in X axis, while "latitude" and "longitude" in the other two axis. It's very important that I need to follow time axis and take the other two numbers and compress it, just like hashing. Any suggestion or ideas, where can I start up with?

推荐答案

车辆的驱动模式也不过是车辆的运动轨迹。比方说，它的三（纬度，长，时间）。既然你要创建一个签名了它我可以建议两种方式：

The driving pattern of a vehicle is nothing but the trajectory of the vehicle. Let's say its a triple (lat, long, time). Since you want to create a signature out of it I can suggest two approaches:

您可以考虑你的整个空间作为一个大立方体（0,0,0）到（M，N，t_max）。 M和N是最大纬度和长RESP，而t_max是要考虑的最长时间。现在，您可以采取一些粒度和划分网格成细和更小的立方体每个小且易于管理。然后，您可以像使用Z顺序曲线（ HTTP一些标准的空间填充曲线：//en.wikipedia。组织/维基/ Z-order_curve ）或希尔伯特曲线（ http://en.wikipedia.org /维基/ Hilbert_curve ）到你的3维数据映射到这仅仅是一个线性排序或签名的单一维度。

You can consider your entire space as a big cuboid (0,0,0) to (M,N,t_max). M and N are the max lat and long resp., while t_max is the maximum time you are considering. You can now take some granularity and divide your grid into fine and smaller cuboids each of which is small and manageable. You can then use some standard space filling curves like Z-order curve (http://en.wikipedia.org/wiki/Z-order_curve) or Hilbert curve (http://en.wikipedia.org/wiki/Hilbert_curve) to map your 3 dimensional data to a single dimension which is just a linear ordering or a "signature".

您可以先线性化的二维数据（纬度，长）的顺序。一旦你做到了这一点，你可以使用这里提到的分段聚集逼近（PAA）技术：时间序列的符号重新presentation，与启示 流算法（dl.acm.org/citation.cfm?id=882086）。

You can first linearize the ordering of the two dimensional data (lat, long). Once you have done this, you can use the Piecewise Aggregate Approximation (PAA) technique mentioned here: A Symbolic Representation of Time Series, with Implications for Streaming Algorithms (dl.acm.org/citation.cfm?id=882086‎).