Contact information
University:
KULeuven
Team members:
Faculty advisor:
Cheng-Ming Chen
Email address:
Submission language
English
Project information
Title:
2x2 MIMO transmitter
Description:
The goal of this project is to implement a MIMO (multiple input, multiple output) system. To start, a SISO (single input, single output) system was given. The idea of a MIMO system is to have multiple data streams that are transmitted at the same time using the same frequency, bandwidth and modulation. The system that will be described is a 2x2 system, this implies that there are two transmitting antennas and two receiving antennas. By using more antennas and applying the right techniques, a higher throughput can be achieved. This can be used e.g. by sending multiple video streams at the same time. To get this 2x2 MIMO system working, there are a lot of challenges to overcome.

Products:
To implement this project we used LabVIEW Communications for USRP RIO and the 802.11 application framework. The used USRPs are from the type NI USRP-2942R. These tools gave us a good start since the application framework already implemented a fully working 802.11a transmitter. Using labVIEW and the USRPs were very helpfull to test the system since it gave live feedback from the implemented system.
The Solution:
Two antennas
The transmitter, also referred as UE, had to be altered to support a MIMO system. It has been altered in several ways. First of all the transmitter uses two antennas instead of only one. This creates two channels that can potentially be used for transmitting data. In the first versions the two antennas transmitted the same data, so only using one channel. Afterwards the frame structure of the 802.11a standard was altered to make use of the two channels. At this point the data stream can be duplicated, so 2 independent data streams can be transmitted.
To enable the two antennas at the same time, the configuration of the antennas should be performed two times. Once for the first antenna (TX0), one time for the second antenna (TX1). This could be performed by surrounding the configuration process with a loop structure and loop over the two antennas. Now the data should be written to both antennas at the same time. This was done by increasing the size of the internal FIFO that connected the baseband system to the RF system. At the baseband side, the stream was duplicated and merged together to go through the FIFO. At the RF side the FIFO was read and the stream was split into two. Each of these two streams was written to the right I/O port to reach the antennas.
The System had now two working antennas, transmitting the same data. Due to these changes the system had more redundancy. This implies that if one antenna is obstructed, the other one could still send to the receiver. It gives the system a higher diversity.
Frame Structure
The used system is based on the IEEE 802.11a standard. This standard uses a specific frame structure to encode and decode the data. The original frame structure is shown in the following figure:

The frames are used for different purposes:
The original frame structure doesn't support a MIMO system. That's why the frame structure should be altered. To make use of the two channels the receiver should be able to do a channel estimation of both of them. This implies that the transmitter should send out a training packet for both channels. This is implemented in our solution by sending out the L-LTF frames, which are used for the channel estimation, twice, each time on a different antenna. In this way we used time-division multiplexing to send both training fields. The altered frame structure can be seen in the following figure:

After implementation of these frame changes the output of the system was written to the PC and plotted using matlab. The results can be seen in the figure.

Now the receiver can calculate the 2 different channels by using the new L-LTF preambles to estimate the channel. This gives 4 channels estimations.


Afterwards the receiver can split the two channels that are used at the same time using these 4 channel estimations and zero forcing. In this way 2 data streams can be send at the same time, doubling the spectral efficiency.