I am a student of Telecommunications Engineering at the School of Engineering of Seville and I'm developing a GPS spoofer. I'll use the results of this work for my Bachelor thesis. I have been inspired by the related works at The University of Texas at Austin. 
I have used GNU Radio for developing the software modules of the spoofer. Diagram 1 you can see our spoofer sheme for one single satellite. The fake GPS signal is generated using GNU Radio (at 10.230M samples per second) and I transmit it with USRP N200 (at 10M samples per second). I generate the C/A code (+-1), then multiplied by the generated navigation message (+-1) at their corresponding rate.
But when I transmit the signal (delay 0) the receiver receives a small random delay (increase/decrease).
I think it is a problem with the resampled signal (10230000 -> 10000000). You know what might be the fault?
I have spent many weeks in this issue, I hope you can help me.
Thank you very much in advance.
 Because the transmitter does not support a rate of 10.230M samples per second.
 One sample per bit.
 C/A code sample rate (GPS L1) is 1023 Kbits/s --> 97.75 us (1/1023K) each bit. Navigation data sample rate (GPS L1) is 50 bits/s --> 0.02 s (1/50) each bit.
I chose a rate of 10230000 samples/s to multiply by an integer C/A code ( 1/10230000 * 10 = 97.5 us) and navigation data ( 1/10230000 * 204600 = 0.02 s)