|
|
【答案】应助回帖
★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★
unitjava: 金币+15, 翻译EPI+1, ★★★很有帮助, 谢谢，另请看http://emuch.net/bbs/viewthread.php?tid=4320076 2012-03-30 11:20:16
Compare with common data, The data in Internet of things are always massive, distributive, time-relating and location- relating, also, the sources of data are distinctive and the resources of the joint are limited.
all of the features above bring in numerous problems about central data mining.
the enormous numbers in the internet of things need a real-time processing. So a pretty high demand will be needed if central construction is adopted.
Therefore, it is not a viable strategy to put all the relevant data together considering some aspects on data like, safety, privacy fault-tolerance, commercial competition, restriction of law and others.
The resources of the panel joints are limited. It doesn’t optimize the transition of the expensive resources to lay all the data on the central joint. In most cases, the central joint doesn’t need all the data, but need to estimate some parameters, therefore, we can preprocess the original data in the distributive joint, then send necessary information to receivers. |
|
回复此楼