3770 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 9, NO. 12, DECEMBER 2010 A-GNSS Sensitivity for Parallel Acquisition in Asynchronous Cellular Networks Seung-Hyun Kong, Member, IEEE, and Wooseok Nam, Member, IEEE Abstract—Increasing the dwell time in two-dimensional frequency-time hypothesis testing is, in practical terms, one of the most effective ways for Assisted Global Navigation Satellite Systems (A-GNSS) and GNSS receivers to achieve higher sensi- tivity. In an asynchronous cellular network, however, a mobile terminal may have a non-negligible unknown clock drift rate error originating from the received cellular downlink signal. In such a case, increasing the dwell time may not necessarily result in the expected sensitivity improvement. In addition, a mobile terminal in a rich multipath environment may experience jitters in the code phase of the resolved first arrival path due to short- delay multipaths, which also degrades the sensitivity. In this paper, new decision variables using a lone or a pair of adjacent 1 cells for code phase hypothesis testing and clock drift rate hypothesis testing are proposed to cope with the unknown code phase drift rate error and the effect of code phase jitter in a parallel acquisition system. The statistics of the proposed decision variables are analyzed in a Rayleigh fading channel, and the performances of the proposed decision variables are compared with that of the conventional decision variable. Index Terms—Assisted GNSS, assisted GPS, parallel acquisi- tion, asynchronous cellular networks. I. I NTRODUCTION T HE Assisted Global Navigation Satellite System (A- GNSS) is one of the most popularly used mobile localiza- tion technologies to support the growing demand for location- based services. A-GNSS can achieve high sensitivity with assistance from cellular networks, as defined in [1] and [2]. An A-GNSS receiver performs two-dimensional frequency- time hypothesis testing, which can be simply implemented with a parallel acquisition system [3], [4]. In addition, the dwell time can be increased by the accumulation of multiple integration results. For example, the frequency uncertainty of the Global Positioning System (GPS) 1 carrier frequency (1.575GHz) due to the relative motion between a GPS receiver and a GPS satellite is typically within [−6, 6]kHz, and the code phase (time) uncertainty can be as large as the PRN code length (1023 chips) of the GPS 1 Coarse Acquisition (C/A) signal [5]. Typical two-dimensional frequency-time hypothesis testing consists of an outer loop for frequency searches with a step size of less than 500Hz, and an inner loop for code phase searches with a step size of half a chip, as in [5] and [6]. To increase the sensitivity in GNSS signal acquisition, an A- GNSS receiver in a synchronous cellular network, such as IS- 95 and CDMA2000, is provided with an accurate code phase Manuscript received January 15, 2010; revised June 30, 2010 and Septem- ber 3, 2010; accepted September 4, 2010. The associate editor coordinating the review of this paper and approving it for publication was G. Colavolpe. The authors are with the Department of Aerospace Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Korea (e-mail: skong@kaist.ac.kr, wsnam@stein.kaist.ac.kr). Digital Object Identifier 10.1109/TWC.2010.092810.100070 estimate with a few chips (microseconds) of uncertainty and other navigation information by the cellular network [1]. How- ever, the code phase estimate provided to A-GNSS receivers by asynchronous cellular networks, such as the Global System for Mobile (GSM) and the Wideband Code Division Multiple Access (WCDMA) systems, has seconds of uncertainty when the networks do not have Location Measurement Unit (LMU) [2]. Consequently, the acquisition performance and the achiev- able sensitivity in asynchronous networks are poorer than those in synchronous networks. In addition, given that mobile terminals rely on the received downlink signal frequency and timing reference, non-negligible and time-varying clock drift can arise in the receiver-generated clock in asynchronous cellular networks. In practice, stringent carrier frequency and clock drift rate tolerance requirements for the IS-95 and CDMA2000 downlinks are well satisfied, as the base stations are equipped with fine GNSS receivers for the synchronization of all base stations. However, GSM and WCDMA do not have such stringent frequency and clock drift rate tolerance require- ments for downlinks; GSM and WCDMA allow a maximum of ±0.1ppm error in the downlink carrier frequency. Moreover, user terminals are required to maintain synchronization to the carrier frequency of the received downlink signal within ±0.1ppm [7]-[9]. In many A-GNSS applications, indoor environments and deep urban canyons are the most challenging environments, as GNSS receivers in these type of areas usually experience very weak received signal strength. In such challenging environ- ments, GNSS signals often arrive at the receiver via multiple Non-Line of Sight (NLOS) paths and therefore experience considerable path loss due to obstacles such as buildings. These multipaths cause fading in the signal amplitude and entail unknown pseudorange error as well. In [5], it was found that short-delay multipaths constituting a resolved first arrival path can lead to detection errors in code phase. The detected code phase error also varies according to the relative phase and delay between the dominant path and the short-delay multi- paths. Intuitively, when only NLOS multipaths constitute the resolved path, the detected code phase can vary significantly as the delay and phase of each short-delay multipath vary randomly. This results in a time-varying code phase jitter that is not negligible for the GNSS signal acquisition system. Many sensitivity improvement schemes for GNSS receivers introduced in the literature focus on mitigating the large time uncertainty. Recent studies show that GPS navigation bit transition and substantial frequency drift larger than half of the frequency search step size during the GPS signal search can be handled [10], [11]. In [12], joint Bayesian estimation of the time and frequency offsets for GPS is considered for improved 1536-1276/10$25.00 c ⃝ 2010 IEEE