Patent US5504780A successfully refreshed from Google Patents.
US5504780A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Adaptive equalizer using self-learning neural network
Assignee: Nytell Software LLC
•
Filing: 1994-01-06
•
Granted: 1996-04-02
•
Expires: 2014-01-06
Abstract
Abstract A channel equalizer is formed using a self-learning neural network. During a training period, the neural network is taught the channel response function. The network is then used to equalize distortions introduced into signals by the channel. The neural network may be a Boltzmann Machine type of neural network comprising neurons arranged in an input layer, a hidden layer, and an output layer. The neurons are interconnected by bidirectional symmetric weighted synapses. Each neuron is preferably implemented by an analog integrated circuit. Direct communication between the input and output layers helps in faster channel acquisition. The scheme can very easily be extended to multilevel and multisymbol modulation schemes such as QAM and PSK.
Inventors
- 👤 Joshua Alspector
- 👤 Timothy X. Brown
- 👤 Anthony Jayakumar
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-10 01:26
CPC Classifications
G; G06; G06N; G06N3/00; G06N3/02; G06N3/06; G06N3/063; G06N3/065; H; H04; H04L; H04L25/00; H04L25/02; H04L25/03; H04L25/03006; H04L25/03165; H04L2025/03433; H04L2025/03439; H04L2025/03445; H04L2025/03464
Claims Analysis
Total Claims: 15
Claim 1
Independent Claim
110 words
A communications system comprising a transmitter for generating signals, a channel for transmitting said signals and for introducing distortions into said signals, an equalizer for receiving distorted signals from the channel and for equalizing the distortions, and a receiver for receiving equalized signals from the equalizer, said equalizer comprising a self-learning neural network having an input layer of neurons for receiving said distorted signals from the channel and an output layer of neurons for outputting said equalized signals, each of said neurons comprising a node for obtaining a net input current, a current normalizer for normalizing the net input current, a current-to-voltage converter, and a gain controlled cascode output amplifier.
Claim 2
Dependent Claim
110 words
A communications system comprising a transmitter for generating signals, a channel for transmitting said signals and for introducing distortions into said signals, an equalizer for receiving distorted signals from the channel and for equalizing the distortions, and a receiver for receiving equalized signals from the equalizer, said equalizer comprising a self-learning neural network having an input layer of neurons for receiving said distorted signals from the channel and an output layer of neurons for outputting said equalized signals, each of said neurons comprising a node for obtaining a net input current, a current normalizer for normalizing the net input current, a current-to-voltage converter, and a gain controlled cascode output amplifier.
Claim 3
Dependent Claim
39 words
The system of claim 1 wherein said neural network is a Boltzmann Machine type neural network and said neural network comprises a hidden layer of neurons connected between said input layer of neurons and said output layer of neurons.
Claim 4
Dependent Claim
18 words
The system of claim 1 wherein said neural network comprises at least a portion of a VLSI chip.
Claim 5
Dependent Claim
17 words
The system of claim 1 wherein at least a portion of said channel goes through the atmosphere.
Claim 6
Dependent Claim
19 words
The systems of claim 4 wherein said equalizer is located in a portable terminal in a personal communication network.
Claim 7
Dependent Claim
20 words
The system of claim 1 wherein said cascode output amplifier includes means for receiving a noise signal for simulated annealing.
Claim 8
Dependent Claim
28 words
The system of claim 1 wherein said cascode output amplifier includes means for varying a gain so that said network can be settled using the Mean Field Approximation.
Claim 9
Dependent Claim
93 words
An equalizer for equalizing a communications channel comprising means for receiving distorted signals transmitted via said channel, and a neural network for equalizing said distorted signals, said neural network comprising an input layer of neurons, a hidden layer of neurons, and an output layer of neurons, said neurons being connected by symmetric synapses, each of said neurons comprising an input node for obtaining a net input current, a normalizer for normalizing the net input current, a current-to-voltage converter for converting said normalized net input current into a voltage, and an output amplifier stage.
Claim 10
Dependent Claim
93 words
An equalizer for equalizing a communications channel comprising means for receiving distorted signals transmitted via said channel, and a neural network for equalizing said distorted signals, said neural network comprising an input layer of neurons, a hidden layer of neurons, and an output layer of neurons, said neurons being connected by symmetric synapses, each of said neurons comprising an input node for obtaining a net input current, a normalizer for normalizing the net input current, a current-to-voltage converter for converting said normalized net input current into a voltage, and an output amplifier stage.
Claim 11
Dependent Claim
16 words
The equalizer of claim 8 wherein at least a portion of said channel comprises the atmosphere.
Claim 12
Dependent Claim
19 words
The equalizer of claim 8 wherein there is a direct connection between the input layer and the output layer.
Claim 13
Dependent Claim
70 words
An equalizer for equalizing a communications channel comprising means for receiving distorted signals via said channel, and a self-learning neural network for equalizing said distorted signals comprising an array of neurons connected by synapses, each of said neurons comprising an analog integrated VLSI circuit which comprises a current normalizer for normalizing a net input current, a current-to-voltage converter connected to the normalizer, and a cascode amplifier connected to said converter.
Claim 14
Dependent Claim
70 words
An equalizer for equalizing a communications channel comprising means for receiving distorted signals via said channel, and a self-learning neural network for equalizing said distorted signals comprising an array of neurons connected by synapses, each of said neurons comprising an analog integrated VLSI circuit which comprises a current normalizer for normalizing a net input current, a current-to-voltage converter connected to the normalizer, and a cascode amplifier connected to said converter.
Claim 15
Dependent Claim
16 words
The equalizer of claim 11 wherein said neural network is a Boltzmann Machine type neural network.
References Cited (Backward Citations)
US4874963A
Cited
US5083285A
Cited
US5187680A
Cited
US5396581A
Cited
US5272723A
Cited
US5361327A
Cited
US5336937A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1999-04-02 | 1999-10-02 | 2000-04-02 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2003-04-02 | 2003-10-02 | 2004-04-02 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2007-04-02 | 2007-10-02 | 2008-04-02 | $7700.00 | Window Closed / Lapsed |