[Discuss-gnuradio] Maximum possible transmission rate through the benchmark code

[Nazmul Islam]

Hello,

I am running the benchmark_tx and benchmark_rx codes of the narrowband and ofdm folder (gr-digital/examples) between two USRP2's. I have several questions regarding the maximum possible transmission rate in these two codes. I am listing the questions and my experiment details below:

USRP2 with GB Ethernet cable
Daughterboard: RFX400 (on the transmitter side) and FLEX400 (on the receiver side)
GNUradio version: latest (downloaded on May 21, 2012)
Ubuntu: 12.04


Question 1: I ran the benchmark_tx.py OFDM code at 20 MHz bandwidth at the transmitter side and uhd_fft.py at the receiver side. The spectrum analyzer of the uhd_fft could correctly (approximately) estimate the power level of the received signal in the frequency domain. (I used a real spectrum analyzer to check it). I read in the GNUradio tutorials that the FPGA down-converts 
the signal to 8 Mega complex samples per second for USB 2.0 compatibility. Then how could I observe a signal with 20 MHz bandwidth in the uhd_fft spectrum analyzer? I used the following commands:

./benchmark_tx.py -f 450M -W 25M --occupied-tones 400 -M 20M

./uhd_fft -s 25M

Question 2: When I was trying to receive the packets that were sent with 20 MHz bandwidth, I saw overrun in the receiver. I assume that the receiver side computer is not fast enough to keep up with 20 MHz bandwidth. Then how could it detect the strength of the signal in the entire 20 MHz region? Can I record and store the received signal (with 20 MHz bandwidth) in the time domain just using the gnuradio softwares? (i.e. without doing any FPGA programming)


Question 3: I tried to run the benchmark_tx.py code of the narrowband folder at 20 Mbit/sec. But the code was showing underrun whenever the bit rate exceeded 6 Mbit/sec. I used the following command:

./benchmark_tx.py -f 450M -r 7M -M 20

The ofdm transmitter code is keeping up with high bandwidth but the narrowband tx code is not. I just wonder what the reason might be. 


Feedback on any of the questions will be highly appreciated.  Thanks for reading the email.

Nazmul


-- 
Muhammad Nazmul Islam

Graduate Student
Electrical & Computer Engineering
Wireless Information & Networking Laboratory
Rutgers, USA.