<FONT face=Verdana>An algorithm for effective channel order estimation (NECOE) is proposed.Based on subspace decomposition

the space spanned by the eigenvectors of the autocorrelation matrix of the measurement vector can be divided into “signal” subspace and “noise” subspace

a special matrix Gv is constructed by “noise” subspace

and a new theorem is derived and proved

which indicates that the special matrix has only one zerovalued eigenvalue

if and only if La is equal to the effective channel order.Based on the new theorem

the channel order is estimated by modifying the dimension of “noise” subspace and judging the singularity of matrix Gv.Simulations results show that the novel algorithm has perfect performance for low SNRs

especially when the channel impulse response has small head and tail taps.