Adaptive Signal Processing Based on High-Order Statistics

本研究正在从统计文献中探索新的结果，这些结果有望对信号模型的性能和准确性有显著的优势。以前，由于传统的二阶统计量限制，由于隐式高斯假设，只能从观测数据中识别出最小相位（或对称）模型。通过使用高阶累积统计量，研究人员能够识别出信号模型真实相位的良好近似值。这在自适应信道均衡、自适应回波消除和自适应阵列处理等应用中具有重要意义。该提案在提交给MIPS司的16个SBIR提案中排名第一。强烈建议为第一阶段提案提供6个月49 499美元的预算经费。

This research is exploring new results from the statistics literature which are expected to have significant advantages for the performance and accuracy of signal models. Previously only minimum - phase (or symmetric) models could be identified from observed data due to an implicit Gaussian assumption through the conventional restriction to second-order statistics. By use of higher-order cumulant statistics, the researcher is able to identify a good approximation to the true phase of the signal model. This will be important in applications to adaptive channel equalization adaptive echo cancellation, and adaptive array processing. This proposal was rated first out the sixteen SBIR proposals submitted to the MIPS Division. Funding in the budgeted amount of $49,499 for 6 months is highly recommended for this Phase I proposal.