Nonlinear Channel Estimation Using Separated Linear Filters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for modeling non-linear communication channels, such as those involving high power amplifiers in satellite transponders, are computationally complex due to the high number of coefficients required by the Volterra model, leading to high estimation and inference complexities.
Innovation Solution
A method is proposed to estimate the non-linear channel by obtaining pilot sequences with specific peak amplitudes, estimating the convolution of linear filters, determining candidates for the linear filters, and selecting the non-linear function, reducing the number of coefficients needed compared to the Volterra model.
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 (linear filter, non-linear function, linear filter), but the number of coefficients becomes very high, leading to high estimation and inference complexity
Solution Approach 1:
The patent segments the non-linear channel into three distinct components: a first linear filter h, a non-linear function c, and a second linear filter g. This segmentation allows each component to be modeled separately with fewer parameters, avoiding the exponential growth of coefficients in the traditional Volterra model while maintaining the ability to represent the complete non-linear channel behavior.
2Device complexity
If the number of coefficients in the Volterra model is reduced, then the computational complexity decreases, but the accuracy of channel estimation may be compromised
Solution Approach 1:
The patent changes the parameters used to describe the channel from the traditional Volterra coefficients to a set of parameters describing the three components (linear filter coefficients, non-linear function parameters, and second linear filter coefficients). This parameter transformation reduces the total number of parameters while maintaining estimation accuracy, as each component can be estimated independently using appropriate methods.
3Measurement precision
If a long training sequence is used to estimate the Volterra model coefficients, then the estimation accuracy improves, but the transmission time and loss of useful data increase
Solution Approach 1:
The patent segments the estimation process into three independent stages corresponding to the three channel components. Each component can be estimated using shorter, targeted training sequences rather than requiring a single long sequence for the complete Volterra model. This segmentation reduces the total training time while maintaining overall estimation accuracy.
4Reliability
If the transmitter uses the complete Volterra model for predistortion, then the compensation accuracy is high, but the computational complexity at the transmitter increases significantly
Solution Approach 1:
The patent segments the predistortion function into three corresponding components that mirror the channel structure. This allows the transmitter to apply predistortion in a structured manner with reduced computational complexity, as each component can be processed separately rather than requiring evaluation of the complete high-dimensional Volterra model.
Solution Approach 2:
The patent enables preliminary determination of the three channel components during the estimation phase, so that the predistortion function can be pre-computed and stored. This preliminary action reduces the real-time computational burden at the transmitter during actual signal transmission.
Data Source
Figure 1~3
Figure 4A
Figure 4B
AI summary
A method in a communication system comprising a transmitter and a receiver communicating through a communication channel comprising a high power amplifier and being modeled as a succession of a first linear filter, a non-linear function and a second linear filter is disclosed. The method comprises : - Obtaining (S40) a first received sequence corresponding to the transmission of a first pilot sequence by the transmitter to the receiver ; - Estimating (S42) r responsive to the first pilot sequence and the first received sequence, r being equal to the convolution of the first and the second linear filters ; - Determining (S44) a plurality of candidates for the first linear filter responsive to the estimated r̂ ; - Obtaining (S45) a second received sequence corresponding to the transmission of a second pilot sequence by the transmitter to the receiver ; and - Determining (S46) 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.