以粗估星曆增進GPS訊號擷取效率

Abstract

當GPS技術整合到手機或手持裝置時，在大部分的情況下，可能都不是熱開機的狀態(甚至可能都是在冷開機狀態下使用)，我們需要一個更有效之訊號擷取方法，以降低剛開機時定位所需要的時間。 本論文利用粗估星曆(Almanac)解算衛星位置及速度，進而預測接收機的可視衛星群，及各別衛星的都卜勒頻移，再根據上次的定位紀錄，統計出該台接收機合理的平均鐘飄頻移，其中粗估星曆的來源，或由該接收機先前之儲存，或由手機之無線通訊服務系統提供。利用上述之輔助資訊，當GPS接收機需要冷開機時，我們即可減少訊號擷取所需搜尋的次數，進而降低冷開機所需的時間。上述之方法亦可應用到微弱訊號擷取演算法，此微弱訊號演算法對10ms的資料做同調積分，而10ms的資料其相對應之頻率解析度為100Hz。 接著，利用SE4110L Front-end接收衛星訊號，並透過實驗證實提出的想法。由實驗結果可發現，頻率預測法可使得訊號擷取時所需的頻率搜尋範圍由原本的±5KHz縮小為±250Hz，而頻率搜尋間隔也可由500Hz縮小至250Hz，搜尋次數原本需要21次減少為3次，因此執行速度也更快。應用到微弱訊號搜尋演算法後，若為靜態之接收機，此方法能有效的將頻移範圍縮小至± 150 Hz 內，若為低速載具，也可利用微幅調整頻率搜尋範圍，在最快的執行速度下，達到對微弱訊號擷取的目標。最後，根據定位紀錄利用最小平方法統計出接收機的平均鐘飄頻移，並討論了鐘飄頻移對微弱訊號擷取的影響。

When GPS technique is going to be integrated with cell phone or hand-held devices, it is expected that in most situations the receiver will be in a cold-start condition. In view of this, a signal acquisition method that can decrease TTFT (Time to First Fix) efficiently is well worth a further study. In this thesis, we propose a method that is able to efficiently reduce the time to acquire the GPS signal by utilizing the almanac data of the satellites. If a user's rough location is known, the aforementioned information (satellite’s almanac data) can be used to determine the visible satellites and their corresponding Doppler frequency shifts. The almanac data can be either obtained from last received (complete) navigation data or transmitted from the cell phone wireless service provider. Another factor that may influence the frequency shift is caused by the drift rate of the receiver clock. In this thesis, we also propose a method to estimate the receiver’s clock drift rate to improve the overall performance of our algorithm, where the quantity of the drift rate can be obtained from the previous navigation records. These two procedures can be used to reduce the number of required searches in a GPS signal acquisition algorithm, and therefore the execution time can be decreased. Since the time needed for acquisition is saved, we can spend much more effort to acquire a relative weak signal. We also apply the proposed algorithm to acquire a weak signal for 10ms coherent integration time. In this case, the down-converted frequency will be processed with a bandwidth of ±50 Hz because 10 ms of data would have a corresponding frequency resolution of 100Hz (1/10 ms). Several experiments are conducted to prove the proposed method. In those experiments, the used data is actual raw GPS signal received by using an SE4110L ASIC-based Front-end. From the experimental results, the follow statements may be concluded. Firstly, the frequency search interval is reduced from ±5KHz to ±250Hz when the frequency shift prediction is applied. In other words, the number of frequency searches is reduced from 21 to 3. Secondly, for a static GPS receiver, the interval of frequency search in the frequency shift prediction algorithm is only ±150 Hz. In view of this, by utilizing the frequency shift prediction algorithm, weak signal acquisition may provide better performance and spend less execution time even in the case of a moving vehicle. Finally, the frequency shift caused by the drift rate is obtained from the previous navigation records by applying least-squares method and the effect of this quantity in weak signal acquisition is also discussed.