Nonlinear Channel Estimation Using Segmented Filter Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for modeling non-linear channels in communication systems, such as satellite transponders, using the Volterra model result in high computational complexity due to a large number of coefficients, making estimation and inference stages inefficient.
Innovation Solution
A method is introduced to estimate the non-linear channel by determining a first and second linear filter and a non-linear function using pilot sequences, reducing the number of coefficients required, thereby decreasing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the Volterra model is used to model the non-linear channel, then the model directly represents the result of the three elements (first linear filter, non-linear function, second linear filter), but the number of coefficients becomes very high (greater than L2*(L1)^K), increasing estimation and inference complexity
Solution Approach 1:
The patent segments the non-linear channel model into three distinct elements: a first linear filter h, a non-linear function c, and a second linear filter g. This segmentation allows each element to be estimated separately using different pilot sequences, avoiding the need to estimate a large number of coefficients in a monolithic Volterra model. The segmentation principle directly addresses the contradiction by reducing the overall complexity while maintaining the representational capability of the model.
2Measurement precision
If a high number of coefficients are used in the Volterra model, then the model can represent the non-linear channel accurately, but the estimation stage requires inverting a large matrix and long training sequences, and the inference stage involves high computational complexity
Solution Approach 1:
By segmenting the model into three estimable elements, the patent enables efficient estimation of each element separately using standard techniques, avoiding the computational burden of inverting large matrices required by the Volterra model while maintaining channel estimation accuracy.
Solution Approach 2:
The patent changes the parameters being estimated from the large number of Volterra coefficients to the smaller set of parameters defining the three separate elements (first linear filter coefficients, non-linear function parameters, second linear filter coefficients). This parameter transformation reduces computational complexity while preserving the ability to accurately represent the non-linear channel.
3Reliability
If the first linear filter output has high peak amplitude to drive the non-linear function, then the non-linear effects are more pronounced and easier to estimate, but the signal may saturate the high power amplifier and cause distortion
Solution Approach 1:
The patent uses two different pilot sequences: a first pilot sequence with low peak amplitude that avoids amplifier saturation while still allowing estimation of the linear filters, and a second pilot sequence with higher amplitude that excites the non-linear effects sufficiently for estimation. This partial action approach allows the system to operate in different regimes for different estimation purposes without causing harmful saturation.
Solution Approach 2:
The first linear filter acts as an intermediary that shapes the pilot sequence before it reaches the non-linear function. By carefully designing this filter and its output characteristics, the system can control the amplitude of signals entering the non-linear region while still obtaining sufficient excitation for accurate estimation, thereby mediating between the need for observable non-linear effects and the need to avoid saturation.
Data Source
AI summary
A method in a communication system comprising a transmitter and a receiver communicating through a communication channel comprises:Obtaining a first received sequence corresponding to the transmission of a first pilot sequence by the transmitter to the receiver;Estimating r responsive to the first pilot sequence and the first received sequence, r being equal to the convolution of a first linear filter and a second linear filter;Determining a plurality of candidates for the first linear filter responsive to the estimated {circumflex over (r)};Obtaining a second received sequence corresponding to the transmission of a second pilot sequence by the transmitter to the receiver; andDetermining the non-linear function and selecting one candidate for the first linear filter among the plurality of candidates responsive to the second pilot sequence and the second received sequence.