Low Complexity Digital Filter for Chromatic Dispersion Compensation in Time Domain

Main advantages of its use

The currently available techniques for chromatic dispersion equalization are given both in the time and frequency domains. Given the available techniques, the main advantages of the present technique can be summarized as follows:

1. Multiplierless implementation, based only on shift and addition operations;
2. Low complexity, power consumption and latency for hardware implementation;
3. Sample-wise processing, resulting in an efficient interface with the remaining DSP blocks.
4. Reconfigurable and flexible architecture for different performance and complexity.

Specifications

The chromatic dispersion found in communication systems based on optic fibre strongly limits the range and output of the signal. In a large part of the optical connections currently installed the compensation of chromatic dispersion is carried out in the optical domain by adding dispersion compensation fibre sections with several disadvantages.

The developed method developed in a Portuguese university and presented here is based on the analytical decomposition of FIR filter coefficients in discrete values, given in powers of 2. This decomposition allows the exploration of the coefficients replication, and taking advantage of the distributive property of multiplication over the addition, significantly reducing the number of multiplications required for the FIR filter. Furthermore, the coefficients decomposition in powers of 2 allows to take advantage of a signed digit (SD) representation to decompose these values in terms of shift and addition operations. Thus, achieving a novel and efficient distributive architecture, which is multiplier free.

This method has its main application in high-speed coherent optical systems, being able to be applied for long haul, core or metro networks.

Applications

Integrate the proposed technology in the digital signal processing systems used in coherent optical receivers.

Intellectual property status

Pending patent.

Current development status

This method has been already tested using Matlab simulations and experimental data, where a PM-QPSK signal operating at a rate of 100 Gb/s is propagated over different fiber transmission length. This method has been validated through the FPGA implementation in offline mode using software tool ISE (Integrated Synthesis Environment) Xilinx, and in real-time mode using a FPGA platform (Xilinx Virtex-6 FPGA.

The obtained experimental results are quite favorable, demonstrating that applying the distributive FIR filter architecture for low values of Δ, it is possible to significantly reduce the complexity of the algorithm without significantly compromising the equalization performance.

Desired business relationship

The university intends to license the technology to companies in the field of optical telecommunications. The university is also available to participate in the development and adaptation of the solution.