An Efficient Acquisition Method for the CSK Signal of QZSS LEX

K. Nakakuki, R. Hirokawa

Abstract:	Quasi-Zenith Satellite System (QZSS) has the LEX (L-band experiment) signal which can realize autonomous centimeter-class positioning with its broadcasting data. The LEX baseband signal is modulated by BPSK(5) that is to say its code chipping rate is 5.115 MChip/s. The signal has a dual code structure with a short and a long code interleaved chip by chip, each having a 2.5575 MChip/s bit streams. The short code is modulated by the Reed-Solomon (RS) encoded navigation messages with code shift keying (CSK). It uses RS(255,223) with 8 bits symbol, which can realize 2,000 bps data transmission. The long code is 410 ms pseudo-random noise (PRN) sequence modulated by a square wave with a period of 820 ms. The short and long code represents data and pilot channel respectively. The short code realizes high data transmission rate compared to other GNSS signals by scarifying a receiver’s architectural simplicity. Modulation of the short code with an 8 bit symbol means circular phase shift occurs every 4 ms in its sequence. Consequently circularity of the code is spoiled which makes it difficult to use frequency domain correlation method such as circular correlation. Furthermore comparing to a GPS L1 C/A code receiver, a symbol decision procedure is required additionally when tracking the short code. This paper will discuss unaided signal acquisition methods in the frequency domain to overcome such difficulties. In order to make its receiver widely used, reducing the time to get navigation messages is the key. Reducing the cost and the size of a receiver is also important. The efficiency of the signal processing method is significant in that sense too. That’s why this paper seeks a fast and efficient method in the first place. Algorithms are verified with our QZSS LEX software defined radio (SDR) receiver. QZSS LEX signal was acquired in an I/Q complex format, i.e. in-phase and quadrature-phase, with a wide-band RF frontend in a sampling frequency of 26 MHz. To acquire the LEX signal quickly in a benign environment where carrier to noise ratio is high enough, acquiring the short code first is the best practice. In this case zeros are used instead of the long code in the locally generated code replica. Pre-detection integration time (PIT) was chosen to be 4 ms, which equals to 104,000 samples of the incoming signal. When adopting a fast Fourier transform (FFT) based frequency domain correlation method for a CSK code, there exists some complexities due to its collapsed circularity of the code pattern. With the PIT of 4 ms, there could be one or two correlation peak after operation of the circular correlation. If the two peaks are authentic, their phase difference is within certain range because the amount of phase shift by a symbol varies only from 0 to 255 chips. From another point of view, if the two peaks are apart from each other more than certain amount of range, one of the peak or both of them are not genuine. In this way you can check those peaks if it’s true one or not. Anyway, if you find a correlation peak, one peak or two, the short code boundary which is aligned to integer multiples of 4 ms can almost be detected. Almost, in this case, means there still be uncertainty due to the phase shift by the symbol. To decide the exact short code boundary, you have to either acquire the long code or decide the absolute symbol value by synchronizing the preamble patter of the symbols. After the short code acquisition the long code can be acquired efficiently during the subsequent pull-in and tracking phase. Symbol values must be specified every 4 ms during those phases and that can be done either by a matched filter or a circular correlation method. During that process, long code can be detected simultaneously by interleaving properly shifted long code with the sort code. The amount of shift can be decided not to overlap the short code correlation peak but to gain as much correlation power as possible. If the circular correlation calculation can be done every 4 ms, the long code supposed to be found within 410 ms. Navigation data can be decoded after the long code acquisition because the absolute values of the symbols can be decided. In this way, if the signal to noise ratio of the LEX signal is high enough, the assumption of which is considered as proper in the centimeter class positioning applications, efficient signal acquisition can be realized and navigation data decoding can be started quickly. Although the method was developed in a QZSS LEX SDR receiver, the result enables us to understand requirements for a hardware implementation and its expected performance. In the above example, it is required to perform a circular correlation with 104,000 samples of data in every 4 ms in each channel. This could be too much calculation load for a small and inexpensive signal processing device. To alleviate this processing load, this paper also seeks other FFT based correlation method such as double-block zero padding (DBZP) and discusses advantages or limitations of each method.
Published in:	Proceedings of the 26th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013) September 16 - 20, 2013 Nashville Convention Center, Nashville, Tennessee Nashville, TN
Pages:	340 - 347
Cite this article:	Nakakuki, K., Hirokawa, R., "An Efficient Acquisition Method for the CSK Signal of QZSS LEX," Proceedings of the 26th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013), Nashville, TN, September 2013, pp. 340-347.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In