 Machine readable label reader system with versatile response selection |
| The invention is designed for an environment in which inexpensive machine-readable label devices ("M... |
|
| Radio frequency-controlled telecommunication device |
| The present invention meets this need by providing telecommunications devices and systems which ... |
|
| Portable computer |
| Accordingly, a portable computer capable of wireless data communication without extra peripheral ... |
|
| Method and apparatus for communications monitoring |
| As will be seen, the present invention satisfies the foregoing criteria. Stated generally, the ... |
|
| Call detail reporting for lawful surveillance |
| OF THE PREFERRED EMBODIMENTS The invention involves compiling and processing call associated data, ... |
|
| Testing device for wireless transmission towers |
| Therefore I claim: 1. In a wireless transmission system having a transmission line, an antenna and ... |
|
| Automotive audio system having active controls in reduced power state |
| Accordingly, it is a principal object of the present invention to provide an audio system and a ... |
|
| Radio communication apparatus and control method therefor |
| An object of the present invention is to provide an improved radio communication apparatus which ... |
|
| Contact image sensor for use with a single ended power supply |
| It is one object of the present invention to provide a contact image sensor for use in a fax ... |
|
| Quality-based handover |
| The present invention uses the concept of a distributed handover to minimize interruption of the ... |
|
|
Neural network radar processor
| Details |
Inventors: Farmer, Michael E.; Jacobs, Craig S.; Cong, Shan;
Assignee: Automotive Systems Laboratory, Inc. (Farmington Hills, MI)
Primary Examiner: Blum; Theodore M.
Assistant Examiner:
Attorney, Agent or Firm: Dinnin & Dunn, P.C.
A neural network radar processor (10) comprises a multilayer perceptron neural network (100.1) comprising an input layer (102), a second layer (122), and at least a third layer (124), wherein each layer has a plurality of nodes (108), and respective subsets of nodes (108) of the second (122) and third (124) layers are interconnected so as to form mutually exclusive subnetworks (120). In-phase and quadrature phase time series from a sampled down-converted FMCW radar signal (19) are applied to the input layer, and the neural network (100) is trained so that the nodes of the output layer (106) are responsive to targets in corresponding range cells, and different subnetworks (120) are responsive to respectively different non-overlapping sets of target ranges. The neural network is trained with signals that are germane to an FMCW radar, including a wide range of target scenarios as well as leakage signals, DC bias signals, and background clutter signals. |
|
DETAILED DESCRIPTION The instant invention overcomes the above-noted problems by providing a system and method of processing a radar signal using a neural network that processes the intermediate frequency in-phase and quadrature phase signals sampled in time from a FMCW radar to detect the range of targets illuminated by the associated radar transmit signal, wherein the result of processing by the neural network is similar to the result from the conventional FMCW signal processing steps of DC bias removal, leakage removal, Fast Fourier Transformation, and CFAR detection.
The associated neural network radar processor can be implemented on a neural network processor chip for reduced cost and improved reliability.
The instant invention also provides a method of training the neural network with signals that are germane to an FMCW radar, including a wide range of target scenarios as well as leakage signals, DC bias signals, and background clutter signals.
In accordance with a first aspect, a neural network radar processor comprises a multilayer perceptron neural network comprising an input layer, a second layer, and at least a third layer, wherein each layer has a plurality of nodes.
Each node of the input layer is operatively connected to every node of the second layer.
The second and third layers comprise a plurality of subsets of nodes, wherein nodes from one subset of the third layer are operatively connected only to nodes of one subset of the second layer, there being a one-to-one correspondence between subsets in the second and third layers.
The respective interconnected subsets constitute respective mutually exclusive subnetworks.
The outputs of the nodes in the third layer are operatively connected to outputs of the neural network, and the neural network is trained so that each output node is responsive to a target at in particular range cell and each subnetwork is responsive to targets within a set of target ranges, wherein different subnetworks are responsive to respectively different non-overlapping sets of target ranges
|
|
Cell Phones 
