FIG. 4G shows the SINR of three users when the precoding matrix is updated based on observed interference channels at the respective receivers, supplemented with an explicit sounding. FIG. 4G shows that the sounding can be substantially reduced when the precoding matrix can be kept more up-to-date with other means. In FIG. 4G, a single sounding is needed to keep the degradation to within 3 dB for the whole duration. In the equivalent FIG. 4D, about 30 soundings are needed to achieve the same. And even then, the SINR in FIG. 4D shows an undesirable “saw-tooth” behavior, while the SINR in FIG. 4G is much smoother as a function of time. The approach of updating the precoding matrices based on crosstalk/interference channels measured at the receiver shows a lot of promise as an alternative to frequent sounding. It may not replace explicit sounding, but when used in combination with explicit sounding it will significantly reduce the needs for explicit channel sounding.
