Читать книгу Position, Navigation, and Timing Technologies in the 21st Century - Группа авторов - Страница 140

There is a growing demand for high‐quality PNT solutions for mobile devices due, on the one hand, to FCC’s E911, which requires mobile telecommunication operators to locate their subscribers in case of emergency [1], and, on the other hand, to the emerging market of location‐based services (LBSs) [2]. Both E911 and LBS inevitably involve positioning in urban and indoor environments, where, unfortunately, GNSS alone cannot provide continuous, reliable, and accurate position solutions to its users. Consequently, the search for alternative technologies to supplement or replace GNSS in these circumstances marches on, as evidenced by the many chapters devoted to this topic in this book. In this regard, this chapter addresses broadcasting signals, often referred to as SOOP, which are not designed for but contain useful information that can help find PNT solutions.

The use of broadcast radio signals for PNT is not new. Since the 1960s, the National Institute of Standards and Technology (NIST) radio station WWVB in Fort Collins, Colorado, has been providing precise time and frequency reference nationwide and is still in use today by many clock radios, wall clocks, and other devices [3]. WWVB continuously broadcasts time and frequency signals at 60 kHz, in the radio spectrum band known as low frequency (LF). The time code of the WWVB signal contains all the information necessary to synchronize radio‐controlled clocks in the United States and surrounding areas. In addition, the carrier frequency of 60 kHz is often used as a reference for calibration of electronic equipment.

An early example of using analog television (TV) signals for navigation is TELENAV [4]. TELENAV receives TV signals from a triad of stations or two pairs of stations at known locations and generates the time differences of arrival (TDOAs) to establish hyperbolic lines of position (LOP). The intersection of LOPs provides the user position [5]. Repetitive waveforms such as the horizontal and vertical sync pulses may be used as reference points for the determination of time delays, but the color burst signal is preferred for the purpose. Deploying TELENAV requires synchronization among received TV signals. It can be achieved when a number of stations across a large area simultaneously transmit the same program in a network broadcasting mode, when one station takes the feed off the air from another station, or when an inter‐station cross time synchronization is established through either a common carrier signal, an off‐the‐air signal, or other means.

Radio signals are broadcast not only by terrestrial transmitters but also by low‐earth orbit (LEO) or geosynchronous satellites as well as from such airborne platforms as blimps. This chapter will focus on terrestrial transmissions of digital television (DTV) or digital video broadcast (DVB) signals. Such radio SOOP are available in populated areas as they are designed primarily for indoor reception, where GNSS often fails. Indeed, DTV and DVB signals, now deployed worldwide [6], offer many advantages over their analog predecessors, which can be exploited for precise timing, positioning, and navigation [7–11].

As discussed in [12], terrestrial transmissions can have high signal power on the order of hundreds to thousands kilowatts (kW), thus covering a large area. The transmission frequency is in the VHF and UHF bands (300–900 MHz), thus resulting in better urban propagation and building penetration than GNSS signals in the L‐band around 1.5 GHz. Besides, DTV transmitters are typically sited on the highest land near inhabited areas with antenna towers stretching several hundred feet above the natural elevation. The line‐of‐sight (LOS) transmissions are mostly horizontal, reaching indoors via windows or through walls, with much easier access than for GNSS signals, which come down across roofs. Therefore, good reception is expected indoors. Since the DTV transmitters are fixed on the ground, DTV signals experience less Doppler frequency shift than GNSS signals from orbiting satellites. Tracking loop bandwidth can be tuned down in favor of noise performance over dynamics. The bandwidth of terrestrial DTV signals is between 6 and 8 MHz, wider than that of GPS C/A‐code and comparable to the chipping rate of GPS P(Y)‐code. Wider bandwidth leads to more accurate timing and ranging, which in turn leads to better positioning. Annual, seasonal, and diurnal changes result in variations in propagation delay over the radio path length. However, at DTV frequencies and over a short distance of several hundred kilometers, the delay is much less than that of ionosphere and troposphere propagation delay experienced by GNSS signals. The last – but not the least – point is that broadcasting signals can be used for free as far as the purpose of PNT is concerned since the broadcasting infrastructures already exist, except for the add‐on PNT capability in user devices.

Commonly used mechanisms for obtaining a PNT solution with SOOP [5] include (i) signal power pattern matching (fingerprinting), which requires a pre‐established database or a map containing location‐dependent signal signatures; (ii) triangulation, which requires a means of measuring the angles of arrival (AOAs) of radio signals; (iii) trilateration, which measures the ranges to signal sources either from the received signal strength (RSS) via a propagation loss model or through the time of flight (TOF); (iv) multilateration, which measures differential ranges to a pair of signal sources; (v) dead reckoning, which measures changes in range to radio sources, either from differences in time of arrival (TOA) or changes in carrier phase (Doppler); and (vi) possible combinations thereof. PNT with SOOP can be achieved for a single user or a group of users in a collaborative manner. The latter requires a mobile ad hoc network for data exchange and collaboration among the networked users with possible inter‐nodal ranging [13, 14], which is out of the scope of the present chapter.

To utilize a radio signal as a signal of opportunity for PNT, one needs to know the location of its transmitter either from a database or estimation as well as its transmission time. Besides, the signal must possess recognizable characteristics for identification and for estimation of the times of arrival (TOAs) and/or AOAs at reception. In practice, however, several issues have to be addressed.

 No Timing and Location Coded on Signals. In GNSS, the time of transmit (TOT) and the satellite orbit at transmission can be decoded from the navigation message embedded in the received signals. With SOOP, however, there may be no explicit information about timing and location of transmitter (LOT) modulated on broadcast signals. In fact, one of the major challenges in positioning with SOOP is how to cope with the unknown TOT and LOT.

 Source Locations. As stated above, a critical aspect of positioning with SOOP is how to obtain accurate and up‐to‐date knowledge of the locations of signal sources or a database of location‐dependent signatures. Typically, the source locations such as DTV transmitters can be found from a regulatory registry or by a means of intelligence. A location‐dependent signature database, however, is only available for those regions that have been surveyed beforehand. Although the general characteristics of a SOOP can be learned from the standards to which the signal design adheres or estimated with SIGINT, there are circumstances in which the locations of signal sources are difficult to obtain. Some sources may be easily moved around. It is costly to build and maintain such a database. An alternative is to apply simultaneous localization and mapping of emitting radio sources (SLAMERS) [15, 16].

 Clock Errors. The clocks of SOOP transmitters are initially unknown to a user, each subject to a different bias and drift. Although there are synchronous transmitters such as the single‐frequency network (SFN) for DVB, most of SOOPs are not synchronized. For synchronous transmitters, there is only one clock error between the network and the user, which can be estimated as part of the navigation solution. However, for asynchronous transmitters, there is a clock error term for each transmitter. To solve for such clock unknowns, self‐calibration can be applied when external information is available [17]. A reference station at a known location can estimate the transmitter clock errors and pass it along to the users [18, 19]. In the same way, two collaborative users can form a single spatial difference between their timing measurements with respect to a signal source, thus eliminating the common clock error [20, 21]. Both schemes, however, require a data link and synchronization between them. Another way to remove the clock error relative to a source is to form a temporal difference, leading to differential or relative range measurements for radio dead reckoning [22–25], which can be further combined with other sensors [26–27].

 Number of SOOP and poor geometric dilution of precision (GDOP). The number of “independent” SOOP sources of the same kind in a region is typically not enough for robust and precise position location. Several antennas may be mounted on the same transmission tower. This tends to produce a rather poor GDOP. The problem may be alleviated by using different types of (mixed) SOOP such as TV and cellular signals [29, 30]. Making known displacements, a way to rotate the LOS vectors, is equivalent to adding fictitious sources at different locations to improve GDOP [21, 22]. Taking orthogonal measurements (e.g. obtain AOA in addition to TOA) is another way to enable positioning with limited signal sources.

 2D versus 3D. Due to the limited height of terrestrial transmitters, the position solution is most likely to be two‐dimensional (2D) rather than three‐dimensional (3D). Compensation for slant ranges is required if significant height differences among transmitters and/or users are expected. Such compensation is facilitated when a digital terrain elevation database (DTED) is available. The height can also be solved with a radar or barometric altimeter.

 Multipath at Reception. At reception, severe multipath is expected, particularly in an urban environment [31]. Multipath may create deep fading. Frequency diversity coding (OFDM) and spatial diversity combining (multiple‐input multiple‐output or MIMO) are techniques for better channel equalization. Rapid fading due to motion requires agile and robust tracking of code and carrier. While non‐line‐of‐sight (NLOS) signals are desirable for communications to reach shadowed areas, it is problematic for ranging. NLOS signals can be excluded, de‐weighted, or estimated as bias with robust estimation techniques. On the other hand, multipath can be exploited for constructive use [32, 33]. It can improve the positioning geometry when NLOS paths can be resolved from an environment map. In fact, multipath is troublesome only for geometric positioning methods, but it is rather a blessing for feature‐matching based (non‐geometric) methods, where multipath makes each location rather unique in terms of a rich set of features.

 Signal Integrity/Authenticity. By its very nature, the use of SOOP faces the issues of integrity and authenticity in the context of navigation warfare. The signal physical characteristics and the information content carried on the signals can be used for authentication and assurance. Mixed SOOP and other types of sensors may be used for cross‐checking to ensure source authenticity, measurement integrity, and solution viability.