Software Defined OFDM System for wireless In-Battery Communication

Transcription

1 Software Defined OFDM System for wireless In-Battery Communication Bachelor's thesis of Mateusz Loch Advisor: Dipl.-Ing. Damián Ezequiel Alonso Supervisor: Prof. Dr.-Ing. habil. Klaus Dostert INSTITUTE OF INDUSTRIAL INFORMATION TECHNOLOGY 1 4/30/16 Prof. Max Mustermann - Präsentationstitel KIT Research university in Helmholz Association

2 Structure Motivation Theoretical fundamentals Model description Simulation Measurement Perspectives and conclusion 2

3 Motivation A wireless data transmission for a Battery Management System in an electric vehicle For small frequencies, a channel with a flat fading, PSK can be used For higher frequencies the channel presents a frequency-selective fading, therefore a more robust modulation technique is required 3

4 OFDM a robust modulation technique Modern technique widely used nowadays, e.g. WiFi or LTE Efficient spectrum utilization provides high data rates Good solution against the multipatch propagation The target of my work: implementation of an OFDM system in GNU Radio and selection of its optimal parameters 4

5 Theoretical fundamenals Every OFDM symbol contains a large packet of bits One OFDM symbol consists of a big amount of simple PSK symbols overlapping with each other in time domain All the PSK symbols are on different subcarrier frequencies, orthogonal to each other 5

6 Original model gr-digital examples GNU Radio & USRP's used for the implementation Many scripts and signal processing OFDM blocks already available Combined tx_ofdm and rx_ofdm from gr-digital/examples 6

7 Original model - verification It works only at high SNR Impossible to evaluate its physical level performance 7

8 Original model - disadvantages Model for bursts transmission, modifiable OFDM frame length CRC blocks Complicated Header Loop Header Loop Payload demodulation 8

9 Original model - header loop OFDM frame structure Training seq. 1 Training seq. 2 Data Packet Header Payload 1 Payload 2... OFDM Symbol OFDM Frame Loop funcionality Header & data symbols Header/Payload demultiplexer Payload demodulation 9 Header demodulation Channel estimation, CRC check, packet_len check fileout

10 First, temporal solution add_fl block add_fl (frame_len) still allows old system to function It ignores the CRC-check in the Header 10

11 Schmidl & Cox synchronisation 11

12 Simulation threshold parameter Decreasing the threshold from the default value 0.9 have allowed for a data reception at a lower EbNo On the other hand, it have lead to systematic errors, which can be noticed from the asymptote for a high EbN0 Up from this point, the system can be tested by plotting the BER(EbNo) curves 12

13 Simulation process Transfer bits, Store results Get signal Power Evaluate noise 13 Evaluate BER

14 Evaluation of BER The pseudo noise (PN) sequence is being repeated at the input. The loss of the continuity of the bit stream, due to a failure in the detection of a whole frame in the Schmidl & Cox block, doesn't affect the BER, if only the length of the OFDM frame is equal to the length of an PN sequence. OFDM packet length [bytes] = PN sequence length When choosing the packet length (symbols pro frame) according to above equality, the BER can be separated from a loss of the whole frames. The single bit errors will be visible, rather than bursts of errors 14

15 The influence of packet length Any packet length can be choosen, as far as Packet_len = PN_seq_len The parameters of a single OFDM symbol are set according to the standard a: fft 64, 52 used subcarriers, 4 pilots, 48 data bits Later, the PN sequence of a length 24 is used, to allow a comparison with other IIIT student's systems 15

16 Final version of the OFDM system Keeping in mind the upcoming measurement with the USRP's and a computational demand of a model without the Header Loop the system had been redesigned once more Old New Blocks FFT 2 1 Frame

17 Modified Header/Payload Demux Data symbols Data symbols Header/Payload demultiplexer FFT Hepay demux FFT Channel estimation FFT Channel estimation equalization equalization demodulation demodulation Output Output 17

18 Simulation of different FFT lengths A higher FFT results in a better curve in AWGN The higher the FFT, the synchronization symbols are longer The better robustness comes at the expense of useful data rate, because there are more training sequence bits for every data symbol s. w. 1 s. w. 2 sync word. 1 data data sync word. 2 training seq data data data data training seq. 2 data

19 Simulation of different FFT lengths - QPSK 19

20 Simulation Channel: AWGN Comparison with an OFDM model in Simulink, parameters: -fft32, cp4, -used subcarriers: 32 -modulation qpsk Compared with the theoretical curve: 3 db shift, due to the presence of cyclic prefix and unused subcarriers 20

21 Demonstrator Metal box is a good shield against signals from the outside no interference occurs Data transmission between the master antenna (left) and one of the four slave antennas Two USRP's are connected to the antennas 21

22 Measurement process Transfer & store bits Measure noise Measure signal + noise 22 Calculate BER

23 Measurement FFT 64 used as the least computational demanding method Channel: demonstrator, Helix antenna 2 db shift between BPSK and QPSK measurement is observed 23

24 Outlook Testing the OFDM model with the PIFA antennas Such measurement would require to rebuild the channel prototype and it has not been carried out due to the time constraints Selection of the subcarriers where the CTF is highest 24

25 Conclusion Two functional versions of OFDM flowgraph have been tested The performance of the systems is acceptable for many different parameters, which can be easily set The selection of the optimal parameters for the IntLiIon Project, would only make sense in the PIFA setup - the other single carrier implementations show a better performance in the helix antenna case 25

26 Thank you for your attention 26