M ,(d)–(f)amountoftransmitteddata(TD). Fig.1 .Comparisonofdirectanddeconvolutionapproachestocross-correlationestimationfor:(a)–(c)variabletruncationlength

(1)

102 103 −40 −35 −30 −25 −20 −15 −10 −5 0 5 10 M( s ample s ) N M S E (d B )

dir e c t c r os s -c or r e lat ion s y mme t r ic de c onvolut ion as y mme t r ic de c onvolut ion c ons e ns us de c onvolut ion ( k = 1) c ons e ns us de c onvolut ion ( k = 2) c ons e ns us de c onvolut ion ( k = 3)

(a) SNR = ∞ dB

102 103 −30 −25 −20 −15 −10 −5 0 5 10 M( s ample s ) N M S E (d B )

(b) SNR = 10 dB

102 103 −15 −10 −5 0 5 10 15 20 M( s ample s ) N M S E (d B )

(c) SNR = 0 dB

102 103 −40 −35 −30 −25 −20 −15 −10 −5 0 5 T D ( s ample s ) N M S E (d B )

(d) SNR = ∞ dB

102 103 −30 −25 −20 −15 −10 −5 0 5 T D ( s ample s ) N M S E (d B )

(e) SNR = 10 dB

102 103 −20 −15 −10 −5 0 5 10 15 20 T D ( s ample s ) N M S E (d B )

M ,(d)–(f)amountoftransmitteddata(TD). Fig.1 .Comparisonofdirectanddeconvolutionapproachestocross-correlationestimationfor:(a)–(c)variabletruncationlength

(a) SNR = ∞ dB

(b) SNR = 10 dB

(c) SNR = 0 dB

(d) SNR = ∞ dB

(e) SNR = 10 dB

(f) SNR = 0 dB

Fig. 1

. Comparison of direct and deconvolution approaches to cross-correlation estimation for: (a)–(c) variable truncation length