公开(公告)号：US06445734B1

公开(公告)日：2002-09-03

申请号：US09343873

申请日：1999-06-30

申请人： Li Chen ,  Farhad Aminian ,  Keith T. Chu

发明人： Li Chen ,  Farhad Aminian ,  Keith T. Chu

IPC分类号： H03H730

CPC分类号： H04L25/03038 ,  H04L2025/03707

摘要： Equalizer training is validated by techniques to detect false training. The validation techniques involve checking the unique time domain energy shaping of a plurality of taps of a trained equalizer of a modem for good training. The energy measure for each equalizer tap of the plurality of taps is approximated by summing the magnitude of a real component and a magnitude of an imaginary component of an equalizer coefficient corresponding to the equalizer tap. A tap with a highest approximate energy (i.e., the main tap) is then determined. Next, the main tap energy is compared to a main tap threshold and a predetermined guarding threshold. The plurality of taps are arranged in a time index ordered sequence. In determining the main tap threshold, a first energy sum of a beginning set of taps in the sequence is compared with a second energy sum of a last set of taps in the sequence. The lesser tap energy sum between the first energy sum and the second energy sum is the main tap threshold. The main tap is then compared with the main tap threshold and the predetermined guarding threshold. If the main tap is greater than the main tap threshold and the predetermined guarding threshold, then the train of the equalizer is good. If the main tap is less than the main tap threshold or the predetermined guarding threshold, then the train of the equalizer is false. The above-mentioned validation techniques are effective to recognize the unique time domain energy shaping characteristics of good-trained equalizer taps.

摘要翻译： 均衡器训练通过检测虚假训练的技术进行验证。 验证技术涉及检查用于良好训练的调制解调器的训练均衡器的多个抽头的独特时域能量整形。 通过对实际分量的大小和对应于均衡器抽头的均衡器系数的虚分量的幅度求和来近似多个抽头中的每个均衡器抽头的能量测量。 然后确定具有最高近似能量的抽头（即主抽头）。 接下来，将主抽头能量与主抽头阈值和预定保护阈值进行比较。 多个抽头以时间索引排序顺序排列。 在确定主抽头阈值时，将序列中开始的抽头集合的第一能量和与序列中最后一组抽头的第二能量和进行比较。 第一能量和和第二能量和之间较小的抽头能量之和是主抽头阈值。 然后将主抽头与主抽头阈值和预定保护阈值进行比较。 如果主抽头大于主抽头阈值和预定保护阈值，则均衡器的列是好的。 如果主抽头小于主抽头阈值或预定保护阈值，则均衡器的列是假的。 上述验证技术有效地识别训练有素的均衡器抽头的独特时域能量整形特性。