An efficient block floating point implementation of the LMS algorithm | IEEE Conference Publication | IEEE Xplore