China
Africa
Explore chapters and articles related to this topic
Blind Signal Separation and Blind Deconvolution
Published in Yu Hen Hu, Jenq-Neng Hwang, Handbook of Neural Network Signal Processing, 2018
Blind deconvolution and blind signal separation are clearly related tasks. Blind signal separation attempts to enforce spatial independence of two or more signal streams, and blind deconvolution attempts to enforce temporal independence of a single signal stream. The combination of these two goals yields the problem of multichannel blind deconvolution. In this task, multiple signal streams are processed by a multiple-input–multiple-output filtering system to obtain parallel output signal streams that are approximately independent from sample to sample and from output to output. Multichannel blind deconvolution is similar to spatio-temporal blind signal separation in that multichannel temporal filtering is required; however, the goals of the two problems are different. The multichannel blind deconvolution problem figures prominently in wideband antenna arrays for wireless communications, in which both spatial separation and temporal equalization of the received symbol streams from multiple users are desired.
Direction-of-Arrival Estimation in Mobile Communication Environments
Published in Lal Chand Godara, Handbook of Antennas in Wireless Communications, 2018
Mats Viberg, Thomas Svantesson
Even if the signal waveforms are not known a priori, the structure can be exploited to perform blind signal separation. During the past decade, many algorithms that are capable of separating and equalizing co-channel signals without using any information of the DOAs have been proposed. The signal properties that have been exploited include constant modulus (CM) [1, 46, 68], non-Gaussianity [21, 69, 118], cyclic correlation properties [2, 152] and finite alphabet (FA) structure [104, 132, 133].
Unveiling drivers of sustainability in Chinese transport: an approach based on principal component analysis and neural networks
Published in Transportation Planning and Technology, 2023
Peter Fernandes Wanke, Amir Karbassi Yazdi, Thomas Hanne, Yong Tan
Neural networks are nonlinear statistical data models or decision-making tools. They can be used to find patterns in data or model complex relationships between inputs and outputs. Adaptive control, predictive modeling, and data-driven applications can benefit from neural networks. Deductive reasoning can be used within networks to facilitate self-learning. Artificial neural networks are commonly used for the following tasks: Time series prediction and modeling using function approximation or regression analysis.Recognizing patterns and sequences, detecting novelty, and making sequential decisions.Filtering, clustering, blind signal separation, and compression.