The channel emulator » History » Version 5

SANCHEZ, Eduardo, 03/25/2017 05:53 PM

1 3 SANCHEZ, Eduardo
h1. Hardware architecture
2 3 SANCHEZ, Eduardo
3 3 SANCHEZ, Eduardo
The required devices to support satellite channel emulation are the NI PXIe-8135 and the NI PXIe-5644R. The NI PXIe-8135 is a high-performance Intel Core i7-3610QE processor-based embedded controller which implements the thermal noise and delay effects. It also controls the NI PXIe-5644R which includes the needed radiofrequency functionality to down-convert the received QPSK signal, perform signal sampling in the baseband domain for further processing and regenerate the radiofrequency signal. The front panel of the NI PXIe-5644R can be observed in Figure zzzzz. The radiofrequency input and output provide independent local oscillators, frequency coverage from 65 MHz to 6 GHz, and sampling frequency of up to 80 MHz.
4 3 SANCHEZ, Eduardo
5 5 SANCHEZ, Eduardo
!NI5644_hardware.png!
6 3 SANCHEZ, Eduardo
7 3 SANCHEZ, Eduardo
Due to the fact that the modems operate in L-band (1-2 GHz) transmitting data at 2 Mbauds, the use of NI PXIe-5644R results in an affordable solution for satellite channel emulation. Figure xxx depicts the tasks carried out by the NI devices to process the incoming QPSK signal.
8 3 SANCHEZ, Eduardo
9 3 SANCHEZ, Eduardo
Channel diagram.png
10 3 SANCHEZ, Eduardo
11 3 SANCHEZ, Eduardo
Besides that, the NI PXIe-5644R also provides an integrated LabView programmable FPGA which enables high power processing. In this project, the signal processing is executed in the NI PXIe-8135 instead of the FPGA due to time constraints.
12 3 SANCHEZ, Eduardo
13 3 SANCHEZ, Eduardo
14 2 SANCHEZ, Eduardo
h1. Software architecture
15 2 SANCHEZ, Eduardo
16 1 SANCHEZ, Eduardo
To order to control the vector signal analyzer (VSA) and the vector signal generator (VSG) of the NI PXIe-5644R and perform the signal processing a source code is developed in LabView which is installed in the NI PXIe-8135.
17 2 SANCHEZ, Eduardo
18 2 SANCHEZ, Eduardo
The LabView development is mainly based on the VST Streaming project and the previous QPSK implementation functions such as the noise generation, the delay effect and the visualization tools.
19 2 SANCHEZ, Eduardo
20 2 SANCHEZ, Eduardo
The front panel is illustrated in Figure wwww, the following features are provided
21 2 SANCHEZ, Eduardo
22 2 SANCHEZ, Eduardo
* Eb/No level [dB]
23 2 SANCHEZ, Eduardo
* Time delay [ms]
24 2 SANCHEZ, Eduardo
* Input/Output frequency carrier [GHz]
25 2 SANCHEZ, Eduardo
* Sampling frequency [MHz]
26 2 SANCHEZ, Eduardo
* Power spectral density
27 2 SANCHEZ, Eduardo
* Eye diagram
28 2 SANCHEZ, Eduardo
* Symbol constellation
29 2 SANCHEZ, Eduardo
30 1 SANCHEZ, Eduardo
Front Panel.png
31 1 SANCHEZ, Eduardo
32 2 SANCHEZ, Eduardo
On the other hand, the whole block diagram implementation consist of five major parts which will be explained below
33 2 SANCHEZ, Eduardo
34 2 SANCHEZ, Eduardo
* PXIe-5644R initialization and configuration
35 2 SANCHEZ, Eduardo
* Acquisition of the QPSK signal
36 2 SANCHEZ, Eduardo
* Noise and delay generation
37 2 SANCHEZ, Eduardo
* Streaming of the channel-applied signal
38 2 SANCHEZ, Eduardo
* Visualization and monitoring
39 2 SANCHEZ, Eduardo
40 2 SANCHEZ, Eduardo
h2. PXIe-5644R initialization and configuration
41 2 SANCHEZ, Eduardo
42 1 SANCHEZ, Eduardo
The initialization and configuration of the VST is shown in Figure lllll
43 1 SANCHEZ, Eduardo
44 1 SANCHEZ, Eduardo
Block diagram Configuration.png
45 1 SANCHEZ, Eduardo
46 2 SANCHEZ, Eduardo
h2. Acquisition of the QPSK signal
47 2 SANCHEZ, Eduardo
48 1 SANCHEZ, Eduardo
In Figure qqq, it can be seen how the acquisition loop reads the samples stored in a Direct Memory Access (DMA) buffer. A DMA channel is composed of two FIFO buffers: one on the computer side and another one on the FPGA side. Therefore, in order to recover the QPSK samples, a function block writes data from the FPGA to the DMA buffer and another function block on the computer side reads the data from the target buffer.
49 2 SANCHEZ, Eduardo
50 1 SANCHEZ, Eduardo
Block diagram Input.png
51 2 SANCHEZ, Eduardo
52 1 SANCHEZ, Eduardo
Since the unsigned 16-bit integer format is used to represent the samples, a function converts these samples into complex symbols in order to reuse the functions developed in the previous implementation. 
53 2 SANCHEZ, Eduardo
54 1 SANCHEZ, Eduardo
Once the data is formatted, the samples are added to a FIFO queue. As in the previous LabView development, the queue approach is implemented to connect the main function loops. 
55 2 SANCHEZ, Eduardo
56 2 SANCHEZ, Eduardo
h2. Noise and delay generation
57 2 SANCHEZ, Eduardo
58 1 SANCHEZ, Eduardo
After removing the complex symbols from the queue, the signal power is calculated in order to determine the required noise power for a given Eb/No value that is selected by the user. So that, a noise signal is generated and added to the complex one as shown in Figure www. 
59 2 SANCHEZ, Eduardo
60 1 SANCHEZ, Eduardo
Block diagram Channel.png
61 2 SANCHEZ, Eduardo
62 1 SANCHEZ, Eduardo
In order to create a time delay t_d in the transmission, the noisy samples must wait in the loop for a period of t_w before being delivered to the DMA buffer. As explained later, it must be noticed that the total delay generated by the emulator comprises the required delay as well as the processing time performed in the VST.
63 1 SANCHEZ, Eduardo
Therefore, the waiting time t_w is calculated as follows
64 2 SANCHEZ, Eduardo
65 1 SANCHEZ, Eduardo
t_w = t_d - (t_n - t_a) 
66 2 SANCHEZ, Eduardo
67 1 SANCHEZ, Eduardo
where t_a represents the time when the samples are acquired from the DMA buffer and t_n represents the time when the noisy samples are received in the output loop. The term (t_n - t_a) corresponds to the processing time that must be removed from the required delay. 
68 2 SANCHEZ, Eduardo
69 1 SANCHEZ, Eduardo
Considering the mentioned aspects, the data transmission to the DMA buffer is delayed by t_w once when the first cluster of symbols is received in the output loop queue. The block diagram implementation can be observed in Figure cccc.
70 2 SANCHEZ, Eduardo
71 1 SANCHEZ, Eduardo
Block diagram Noise.png
72 2 SANCHEZ, Eduardo
73 2 SANCHEZ, Eduardo
h2. Streaming of the channel-applied signal
74 2 SANCHEZ, Eduardo
75 1 SANCHEZ, Eduardo
Finally, the channel-applied signal samples are dequeued and written to the DMA buffer in order to be converted from a digital format to an analog signal as shown in Figure yyyy. These samples are interpolated via a DAC, subsequently, the analog signal is up-converted to the required frequency carrier.
76 2 SANCHEZ, Eduardo
77 1 SANCHEZ, Eduardo
Block diagram Output.png
78 2 SANCHEZ, Eduardo
79 2 SANCHEZ, Eduardo
h2. Visualization and monitoring
80 2 SANCHEZ, Eduardo
81 1 SANCHEZ, Eduardo
In order to analyze the transmitted signal, the symbol constellation, the eye diagram and the power spectral density of the QPSK signal are plotted. These loops are executed in parallel to exploit the multiprocessor computing power. As can be observed in Figure rrr, there are two visualization and monitoring loops: one before and another after the channel.
82 2 SANCHEZ, Eduardo
83 1 SANCHEZ, Eduardo
Block diagram Monitor.png