Geraldo Gomes,Pedro Freire,Jaroslaw E. Prilepsky,Sergei K. Turitsyn
Abstract:This paper introduces a new machine learning-assisted chromatic dispersion compensation filter, demonstrating its superior power efficiency compared to conventional FFT-based filters for metro link distances. Validations on FPGA confirmed an energy efficiency gain of up to 63.5\% compared to the standard frequency-domain chromatic dispersion equalizer.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to reduce the energy consumption of the chromatic dispersion compensation (CDC) module in coherent metro - area links. Specifically, the author proposes a time - domain clustering equalizer (TDCE) based on a complex - valued clustering filter, aiming to significantly improve energy efficiency, especially for applications under different fiber lengths.
### Main problems:
1. **Energy consumption problem**: In an optical transmission system, digital signal processing (DSP) usually consumes about 70% of the power, and the chromatic dispersion compensation module is the main source of power consumption. Therefore, reducing its complexity and energy consumption is a key research direction.
2. **Limitations of existing methods**: Although previous studies have proposed methods to reduce mathematical operations (such as chirp filtering and uniform quantization of filter taps), these methods lack hardware implementation to verify their energy efficiency. In addition, other hardware - implementation studies have only evaluated the efficiency under short fiber lengths and have not used standard metrics.
### Solutions:
- **Time - domain clustering equalizer (TDCE)**: The author proposes a new machine - learning - assisted chromatic dispersion compensation filter and verifies its energy efficiency through FPGA implementation. This method reduces the computational complexity by clustering the time - domain filter taps on the complex plane.
- **Hardware implementation and verification**: The paper does not stop at the theoretical level but also implements the algorithm through FPGA and verifies its energy - efficiency advantages under different fiber lengths (up to 640 kilometers at most). The results show that, compared with the traditional fast Fourier transform (FFT) - based frequency - domain dispersion compensator (FDE), TDCE can save up to 63.5% in energy consumption.
### Formula explanations:
- **Chromatic dispersion compensation filter equation**:
\[
g_k = r\exp\left(-j\frac{\pi cT^2D}{\lambda^2}zk^2\right)
\]
where \(z\) is the fiber length, \(k\) is the filter tap index, \(j\) is the imaginary unit, \(D\) is the dispersion coefficient, \(c\) is the speed of light, \(\lambda\) is the carrier wavelength, and \(T\) is the sampling period.
- **FDE complexity**:
\[
C_{\text{FFT}} = N\left(8\beta\log_2(N)+4\frac{N - M + 1}{N}\right)
\]
where \(N\) is the FFT size (block size), \(M\) is the filter size, and \(\beta\) is a constant, which is 1/2 for the Radix - 2 architecture and 3/8 for the Radix - 4 architecture.
Through the above methods, the paper demonstrates the significant advantages of TDCE in terms of energy consumption and complexity, especially the performance improvement in actual hardware implementation.