Good morning everyone. My name is and for the next 15 or so, I will be presenting some of the work that we have been doing for the past months, at IT and IPL in Leiria, Portugal, on the topic of radio channel sounders. So, as the name suggests, this specific presentation will focus on the IF stage of a real time high-resolution channel sounder employed with the sliding correlation technique.
This is the outline for the presentation. I will start with the motivation and objectives behind this work. Then, I approach the sliding correlation principle used in the sounder. I will also present some block diagrams showing the topology of the system, go through the performance assessment of the sounder, show some photos of the complete system… And will end with some conclusions and future work.
So, the motivation behind this work concerns the characterisation and modelling of propagation channels. We want to measure the influence that each channel makes to a certain radiowave frequency, and also its characteristics in terms of multipath and Doppler.
The objective for this work was to develop and test a swept time-delayed cross-correlation RF channel sounder, or STDCC for short, or even sliding correlation. The sounder had to fulfil the following requirements: Ease in changing its parameters, allowing to maximize the performance depending on the measurement scenario; High time resolution for the channel impulse responses obtained (better than 2ns); Ensure a good dynamic range; And to have the capability to measure multipath components in both amplitude and phase, allowing for Doppler spectrum analysis.
Here we have an example of the transmitter and receiver blocks for a sliding correlation sounder. On the transmitter we have a pseudo-noise sequence generator, and then the signal is up-converted and transmitted. On the receiver, the signal acquired by the antenna is down-converted, filtered and amplified, and then it is correlated in a mixer with a replica of the signal created on the transmitter, but, and this is the important part, this PN sequence created on the receiver has a slightly lower chip frequency. This frequency difference makes the sequences slide against each other in time, and only when they are perfectly aligned in time we have a correlation peak at the output. The sliding of the sequences has another interesting property, lets suppose we have 2 multipath components separated by only a few ns. With this sliding principle, the signal at the output will show a separation in the order of ms, for example. We have therefore a time scale dilation.
The PN sequences best suited for this type of sounder are designated in the literature by Maximal Length Linear Shift Register sequences. These sequences can be generated like showed in the figure. These sequences maximize the signal at the output of the correlation (when perfectly aligned), but if they deviate one chip from each other, the output of the correlation will have its minimum value, like showed in the figure. The real spectrum of a PN sequence has this appearance…
This video will show what I’ve been saying. The red sequence is the one generated at the transmitter, and that was attenuated by the channel. The black sequence, which is generated at the receiver, is identical to the red one with the exception that has a slower clock rate. Playing the video we will see that only when they are both perfectly aligned, we will have a correlation peak. When they deviate one chip from each other the correlation will have its minimum.
This video shows a transmitter sequence in red that is comprised of, let’s say, a direct ray and two multipath components, with different delays between each other. When playing the video, we see that each time one of the received sequences aligns with the receiver PN sequence, we will have a correlation peak.
This figure represents the transmitter block diagram of the developed sounder, and I will be focusing on the IF stage components…
Basically it is comprised of a clock generator from manufacturer RFMD, being showed some of its most relevant specifications. This board is responsible for the clock that controls the chip frequency of the PN sequence.
The PN sequence generator is comprised of 2 different components: -> The Data Pattern Generator in which the computer generated PN sequences are stored in volatile memory. ->The AD9739 converts the data sent by the DPG2 to the analogue domain… Next, the sequence is amplified and filtered, up-converted and transmitted to the radio channel.
This is the complete block diagram for the receiver section of the sounder… The signal acquired by the RF stage is down-converted to base-band, and next it enters the IF section of the sounder…
After initial filtering, the signal enters a mixer to be up-converted to 6GHz, so that it is in the frequency range of the I/Q down-converter.
The I/Q down-converter is responsible for the separation of the signal in its phase and quadrature components…
The signals outputted by the I/Q down-converter are filtered and amplified and enter the correlation mixers, so that real-time correlation can be performed with a replica of the transmitted signal but at a slightly slower frequency and represented by the blocks on the right… The correlation output signals, which have frequencies in the order of Hz to kHz, are then filtered and amplified, and finally acquired by a data acquisition card.
So, in order to test the IF stage, we resorted to this setup which consists in the use of splitters and combiners to simulate different multipath components. With this scheme it is possible to test different delays between the components, simply by changing the cables lengths… With the attenuators it is possible to test different signal level combinations associated to each multipath component.
This table presents the main characteristics of the IF stage, when performing the various tests… The most significant variables, that will determine the sounder performance are: -> Chip period; -> PN sequence length; -> The frequency difference between the transmitter and receiver PN sequences…
This slide shows the results for the test which was intended to confirm the sounder time resolution of 2ns. On the left side we see an overall picture of the output which as the three components. The third most delayed component was achieved with a 15m cable. On the right side, it is visible in greater detail the correlation peaks for the first 2 multipath components.
This figure shows an identical test to the previous one, with the difference being that we attenuated significantly the third component, so that we could see where the lower limit for the system was. From this test we see 40dB of dynamic range, but considering that the upper limit was limited to the output power limit and subsequent splitting of the signals, we could expect the effective dynamic range to be around 50dB.
In order to test the full sounder, i.e. IF stage coupled to a RF stage (in this case operating at 18.7GHz), we first resorted to this bench setup which is similar to the one showed previously.
This figure shows a small time frame of the actual measurement, and we can clearly see two correlated path contributions, which actually correspond to the periodic repetition of the PN sequence being transmitted. While doing several bench tests, we saw an overall dynamic range close to 40dB, although it can be a few dB’s higher depending on the sounder parameters (namely the number of bits used to create the PN sequence).
These measurements were achieved by attaching our receiver box to a robot from the robotics department. This robot had a maximum speed of 0.510m/s, and as we can see from the Doppler spectra, the results show a very good agreement. The figure on the left was when the robot was moving away from the transmitter, and the one on the right was when the robot was closing in on the transmitter. The slight difference in the absolute frequency shift can be explained because when the second measurement was done, the robots’ batteries were slightly more depleted.
The lower white box was used with the single purpose of elevating the receiver box to the level of the transmitter box
The developed IF system allows to clearly distinguish multipath components separated by only 2ns… When compared to the frequency sweep method, we see that the real-time sliding correlation shows equivalent performance in time resolution (for this specific test), with the benefit of not requiring a physical connection between transmitter and receiver, and also allowing to analyse Doppler spectra. The possibility to easily modify the PN sequence properties, allows one to adjust the sounder specifications according to the measurement scenario.
It is intended to integrate the developed IF system on several RF subsystems, followed by rigorous testing of the performance characteristics of the sounder, and find the overall dynamic range of the full system. It is also intended to develop a post-processing method to obtain the Doppler spectrum associated to each multipath component. Perform an extensive study on the influence of the several adjustable parameters of the sounder. And finally, housing of the developed IF stage, allowing for an easy relocation over different measurement locations.
Development and Performance Assessment of a Real Time High-Resolution RF Channel Sounder IGARSS 2011 Vancouver – 29 th July 2011 David Ferreira, DL-IT Rafael Caldeirinha, DL-IT | IPL
The developed IF system can successfully distinguish multipath components separated by only 2ns and has a dynamic range around 50dB.
Despite the limitations caused by an RF faulty component, preliminary tests, both in bench as well in an anechoic chamber, show an overall dynamic range for the complete system (IF+RF) of close to 40dB.
By allowing an easy adjustment of the PN sequence properties, either in its length or chip rate, one can adjust the sounder specifications (e.g. dynamic range, time and Doppler resolutions, etc.) to the measurement scenario/geometry.
Integration of the developed IF stage on other RF front-ends, namely (37, 60 and 300 GHz), allowing for high scalability.
Acquisition of a better data acquisition card, in order to improve the maximum detectable object speed through Doppler analysis of the measured PDP (currently limited to a maximum of 120Hz for a maximum discernible delay of 255ns).
Housing of the developed IF stage allowing for an easy relocation of the equipment over different measurement locations.
IGARSS 2011 | Vancouver – 29 th July 2011
IGARSS 2011 Vancouver – 29 th July 2011 Development and Performance Assessment of a Real Time High-Resolution RF Channel Sounder Thank you. E-mails: email@example.com [email_address]