白丝美女被狂躁免费视频网站,500av导航大全精品,yw.193.cnc爆乳尤物未满,97se亚洲综合色区,аⅴ天堂中文在线网官网

Methods and apparatuses for feature-driven machine-to-machine communications

專(zhuān)利號(hào)
US10785681B1
公開(kāi)日期
2020-09-22
申請(qǐng)人
Yiqun Ge; Wuxian Shi; Wen Tong; Qifan Zhang（CN Shenzhen）
發(fā)明人
Yiqun Ge; Wuxian Shi; Wen Tong; Qifan Zhang
IPC分類(lèi)
H04L29/06; H04W28/06; G06K9/62; H04W4/70; G06N3/04; G06N3/08; H04W84/04
技術(shù)領(lǐng)域
may,dnn,arrow,bs,encoder,be,decoder,sensors,feature,sensor
地域： Ottawa

摘要

Methods and apparatuses for feature-driven machine-to-machine communications are described. At a feature encoder, features are extracted from sensed raw information, to generate features that compress the raw information by a compression ratio. The feature encoder implements a probabilistic encoder to generate the features, each feature providing information about a respective probability distribution that each represents one or more aspects of the subject. The probabilistic encoder is designed to provide a compression ratio that satisfies a predetermined physical channel capacity limit for a transmission channel. The features are transmitted over the transmission channel.

說(shuō)明書(shū)

It should be noted that the assigning of sub-channels to features may be different for different sensors. For example, one feature of the observed subject may be well detected by a first sensor, but poorly detected by a second sensor. Accordingly, the quality and importance of that feature may differ between the two sensors. The first sensor may thus assign a robust sub-channel for transmission of that feature, but the second sensor may assign a less robust sub-channel for transmission of the same feature. Each sensor may transmit a respective control message or header to the receiver to inform the receiver about placement of the feature on the different sub-channels.

The above description discloses a machine-learning based approach for designing an encoder DNN and decoder DNN, which is able to account for the effects of the channel, and does not require knowledge about the raw information. The encoder and decoder are both probabilistic, meaning that they encode/decode probabilistic distributions rather than any particular sample from the raw information. The information representation scheme and transmission scheme are selected based on features extracted from the raw information, where the features represent probability distributions. For example, the features may represent Gaussian distributions (or Bernoulli distributions). The transmitted features may be quantized expectation values representing the distributions, and the transmission schemes used for transmission of respective features may be L1 configurations corresponding to noise variance values that match the variance values of the respective features.

權(quán)利要求

1
微信群二維碼
意見(jiàn)反饋