Dear all,
I am doing channel estimation in gnu radio narrow-band. I'm collecting data at the receiver side after the time_recov block. It has been through the gr.firdes.root_raised_cosine in the time_recov. But the plot I get is as the figures in the attachment. It looks like the fluctuation of the channel amplitude has some pattern, it looks like signal pulse shaping in there. But doesn't the gr.firdes.root_raised_cosine works as the rrc matched filter, and already take off the pulse shaping?
Also, when I do experiment, I found out something strange. According to my understanding, the --tx-amplitude on the transmitter benchmark sets the transmission power. So the bigger the amplitude is, the better performance it should be, since the SNR will be bigger. But in actual experiment, it is totally the opposite way! The amplitude of 0.02, or even 0.002 works better than the default 0.25. Has anyone come across the same problems?
Highly appreciate if anyone can give me some suggestions on this! Thanks in advance!!
Ada
I am doing channel estimation in gnu radio narrow-band. I'm collecting data at the receiver side after the time_recov block. It has been through the gr.firdes.root_raised_cosine in the time_recov. But the plot I get is as the figures in the attachment. It looks like the fluctuation of the channel amplitude has some pattern, it looks like signal pulse shaping in there. But doesn't the gr.firdes.root_raised_cosine works as the rrc matched filter, and already take off the pulse shaping?
Also, when I do experiment, I found out something strange. According to my understanding, the --tx-amplitude on the transmitter benchmark sets the transmission power. So the bigger the amplitude is, the better performance it should be, since the SNR will be bigger. But in actual experiment, it is totally the opposite way! The amplitude of 0.02, or even 0.002 works better than the default 0.25. Has anyone come across the same problems?
Highly appreciate if anyone can give me some suggestions on this! Thanks in advance!!
Ada