Mitigating interference in a signal

[0001] This application claims the benefits of U.S. Provisional Application Ser. No. 60/656,668 entitled “Mitigating Interference in a Signal,” filed Feb. 25, 2005, and hereby incorporated by reference.

[0002] The invention relates generally to the field of structured signal receivers and more particularly to a method for mitigating structured interference in a received composite signal by modeling the interference and then subsequently removing the interference from the signal.

[0003] An example of a structured signal receiver is a Global Positioning System (GPS) receiver which is part of the United States' Global Positioning System (GPS), also known as a radio-navigation satellite system (RNSS). The GPS was established by the United States government, and employs a constellation of 24 or more satellites in well-defined orbits at an altitude of approximately 26,500 km. These satellites continually transmit microwave L-band radio signals in two frequency bands centered at 1575.42 MHz and 1227.6 MHz, denoted L1 and L2, respectively. These signals include timing patterns relative to the satellites' onboard precision clocks (which are kept synchronized by ground stations) as well as navigation messages giving the precise orbital positions of the satellites, an ionosphere model, and other useful information. A GPS receiver processes these radio signals to compute ranges to the GPS satellites; and, by triangulating these ranges, the GPS receiver determines its position and its internal clock error.

[0004] GPS's designers assumed that all transmitters would be aboard satellites located at large and relatively constant distances from all user receivers, consequently generating weak but relatively constant signal levels at the receivers. This assumption drove a number of trade-offs in system and satellite transmitter design and continues to influence receiver development even today.

[0005] Despite this assumption, ground-based transmitters (known as PLs, pseudo-satellites, or simply pseudolites) have been used to complement the GPS satellites from the very beginning. In the foreseeable future, PLs may be incorporated in unmanned aerial vehicles (UAVs). A PL transmits a signal with code-phase, carrier-phase, and data components which may or may not have the same timing and format as the satellite signals. A GPS receiver acquires such a PL signal and derives code-phase pseudo-ranges or carrier-phase measurements to be used in a navigation algorithm in substantially the same manner as for a GPS satellite signal. The major differences between a satellite and PL are that a PL typically does not contain a high-accuracy atomic clock and that the PL position must be described using geographical terms rather than orbital elements.

[0006] Precision navigation and landing systems require reliable and highly accurate position, velocity and time (PVT) information that cannot be obtained by standalone GPS. Precision-guided weapons require reliable PVT information to achieve acceptable Circularly Error Probable (CEP) targeting errors. To meet these requirements, additional radio-navigation transmitters are needed. These transmitters can be additional satellites as specified in the Wide Area Augmentation System (WAAS); ground-based PLs as specified in the Local Area Augmentation System (LAAS); ship-based PLs; or, PLs on UAVs loitering in the air above an area of interest. WAAS and LAAS can transmit either correction data (i.e., differential data) or provide additional ranging information. When these transmitters broadcast augmentation signals in the GPS spectrum, additional interference is introduced into the GPS spectrum. This structured interference is like noise to the receiver, degrading the performance and in some cases preventing a receiver from acquiring and tracking the satellites.

[0007] Moreover, the use of PLs violates one of the key assumptions behind the design of GPS. The distance between a user receiver and a PL can range from short to long, so PL signal levels at a receiver can vary significantly. Relatively strong PL signals may overwhelm satellite signals and jam a receiver. Weak PL signals may be too feeble to allow receiver tracking. The challenge associated with this variable range effect is known as the “near-far problem” in wireless communications.

[0008] Equally problematic is structured interference arising from sharing the GPS radio frequency spectrum with other users or from the encroachment of other users' signals on the GPS spectrum. For example, Mobile Satellite Systems (MSS) downlinks, wind profiler radar, space-based radar, ultra-wideband systems, GPS expansion and the European Radio Navigation Satellite System known as Galileo, employ or may use frequencies in and around the GPS spectrum. In another example, GPS jamming technologies may broadcast interference signals in the GPS spectrum. These various RF systems introduce interference into existing RNSS systems either unintentionally or intentionally.

[0009] Another type of interference is self interference, which results when signals from a radio-navigation transmitter interfere with the reception of radio-navigation signals at the receiver. This type of interference often occurs when a RNSS receiver and transmitter are located physically near (or identical to) each other. Self interference is an extreme case of “near-far problem.”

[0010] In conclusion, many types of structured interference exist within the RNSS RF spectrum. It is desirable to have a method and apparatus to identify and remove such wireless interference that degrades or compromises legitimate radio navigation signals. In particular, it is advantageous to reduce or mitigate the near-far problem in radio navigation.

[0011] The following abbreviations are used herein.

[0012] ADC: Analog to Digital Converter

[0013] AFRL: Air Force Research Lab

[0014] C/A code—Coarse/Acquisition or Clear/Acquisition Code

[0015] CDMA—Code Division Multiple Access

[0016] CEP—Circular Error Probable

[0017] DARPA—Defense Advanced Research Projects Agency

[0018] DGPS—Differential GPS

[0019] DLL—Delay Locked Loop

[0020] DOP—Dilution of Precision

[0021] E code—European code

[0022] DSP—Digital Signal Processing

[0023] FLL—Frequency Locked Loop

[0024] FPGA—Field Programmable Gate Array

[0025] GNSS—Global Navigation Satellite System (ICAO definition)

[0026] GPS—Global Positioning System

[0027] I—In phase.

[0028] IF—Intermediate Frequency

[0029] IMU—Inertial Measurement Unit

[0030] INS—Inertial Navigation System

[0031] LAAS—Local Area Augmentation System

[0032] MAS—Multiple Access System

[0033] MFD—Matched Filter Detector

[0034] MSD—Matched Subspace Detector

[0035] MSS—Mobile Satellite System

[0036] NF—Near Far

[0037] NFR—Near Far Resistant

[0038] P(Y) code—Precision (Encrypted) code

[0039] PL—Pseudolite, pseudo-satellite

[0040] PLL—Phase Locked Loop

[0041] PRN—Pseudo Random Noise code, e.g., C/A Gold codes and the P(Y) codes.

[0042] PVT—Position, Velocity, and Time

[0043] Q—Quadrature

[0044] RAIM—Receiver Autonomous Integrity Monitoring

[0045] RF—Radio Frequency

[0046] RNSS—Radio Navigation Satellite System

[0047] ROC—Receiver Operating Characteristic

[0048] SA—Selective Availability

[0049] SNR—Signal to Noise Ratio

[0050] SV—Space Vehicle (e.g., an RNSS satellite)

[0051] VCO—Voltage Controlled Oscillator

[0052] UAV—Unmanned Aerial Vehicle

[0053] UMP—Uniformly Most Powerful

[0054] USAF—United States Air Force

[0055] WAAS—Wide Area Augmentation System

[0056] It is advantageous to define several terms before describing the invention. It should be appreciated that the following definitions are used throughout this application. When the definition of a term departs from the commonly used meaning of the term, the applicant intends to utilize the definition provided below unless otherwise indicated.

[0057] GPS Codes: Each GPS satellite or PL transmits two or more different codes. Such codes typically include a coarse/acquisition (C/A) code and a precision (encrypted) (P(Y)) code. Each C/A-code is a unique sequence of 1023 bits, called chips, which is repeated each millisecond. The duration of each C/A-code chip is about 1 micro-second. The corresponding C/A-code chip width is about 300 m, and the C/A-code chipping rate is 1.023 MHz (or megachips/s (Mcps)). A P-code is a unique segment of an extremely long (≈10¹⁴ chips) PRN sequence. The P(Y)-code chipping rate is 10.23 Mcps, and the P(Y)-code chip width is about 30 m. The shorter P(Y)-code chip width provides greater precision in range measurements than for the C/A-codes. Other codes exist or are planned for the future, both in U.S. and foreign systems; this invention applies to them as well.

[0058] Wireless Signal Model:

[0059] Let a wireless navigation signal y be modeled as follows:
y=Hθ+Sφ+n  (1)

[0060] where

[0061] H is the desired or target signal (a vector) or signals (a matrix);

[0062] θ is a scalar or vector corresponding to the amplitude(s) of the target signal(s) in H;

[0063] S is the known structured interference signal (a vector) or signals (a matrix);

[0064] φ is a scalar or vector corresponding to the amplitude(s) of the interference signal(s) in S;

[0065] n is noise; and,

[0066] H and S are formed by concatenating known signal vectors, for example:
S=[s₁s₂ . . . s_N]  (2)

• • [0067] where s_i is the i^th interference signal.

[0068] In Phase (I) and Quadrature (Q) Components of a Signal refer to the signal components generated by Quadrature Amplitude Modulation techniques, which is a modulation method using both a carrier wave (e.g., the In Phase or I signal component) and a ‘quadrature’ carrier wave that is 90° out of phase with the main carrier (e.g., the Quadrature or Q signal component).

[0069] Near-Far Interference: Commingling of two or more different wireless signals from one or more wireless sources in such a manner that when the commingled signal is received, the stronger signal component(s) (e.g., likely from a source nearer to the receiver) overwhelm(s) the weaker signal components(s) (e.g., likely from a source farther away from the receiver). In particular, the stronger signal component may “leak” into the signal detector (e.g., cross correlation) for the weaker component and thus compromise the accurate detection of the weaker signal component.

[0070] Near-Far (NF) interference can occur both in military and civilian environments and can be from friendly or hostile sources. Friendly sources include PseudoLites (PLs) placed at airports to enhance navigation. A strong PL signal can actually interfere with the receiver's ability to acquire and track the satellites' signals, thereby unintentionally denying the receiver to be used in navigation.

[0071] Hostile interference may be found in the military arena. A hostile force may deploy ground- or air-based PLs with the intent of confusing GPS receivers within an area. Any type of military hardware that uses a GPS receiver is susceptible to this jamming.

[0072] Structured interference is any wireless (e.g., radio) interference source whose signals can be predictively modeled.

[0073] Self-Interference: Wireless signal interference that occurs when a receiver is collocated (i.e., located within a proximity sufficient to induce interference) with a transmitter. Self interference can result when signals transmitted from a radio-navigation transmitter interfere with a radio-navigation signal received on the same antenna used for transmitting. This type of interference often occurs when a receiver doubles as a transmitter. Self interference is an extreme case of “near-far” interference.

[0074] Multipath & Jamming Interference: Smart jamming can be used to cause an active radio-navigation receiver to lock onto legitimate-appearing but false signal(s); the receiver is then slowly drawn off the desired path causing significant PVT errors. Multipath-like interference is the reception, delay, and rebroadcast of radio navigation signals to confuse a navigation system or user. Multipath signals are signals that have an increased geometric path delay due to reflections of the line-of-site signal.

[0075] Higher Order DLL, FLL, PLL: Generally, the order of a phase locked loop (PLL) is 1 higher than the order of the loop filter. If the loop filter is omitted, i.e., if the output of the phase detector directly controls a voltage controlled oscillator (VCO), a first-order PLL is obtained. The term “order” is defined as the exponent of the largest term in the filter polynomial. Since higher-order loop filters offer better noise cancellation, loop filters of order 2 or more are used in critical applications.

[0076] Massively parallel acquisition scheme is a system that can (at least substantially) continuously acquire signals of interest. Its ability to divide the Doppler, phase and code offset search space is only limited by the number of correlators and speed of the processors. In theory such a system could provide the interference modeling parameters to the present invention.

[0077] Navigation Data: GPS transmits a navigation data message which includes a telemetry word, a hand-over word, clock corrections, SV health/accuracy, ephemeris parameters, almanac, ionospheric model and coordinated universal time data.

[0078] Nominal Satellites are satellites operating normally or within their design specifications.

[0079] Processing Channel: A processing channel of a GPS receiver provides the circuitry necessary to process the signal from a GPS transmitter (e.g., a satellite, or pseudolite). The acquisition and tracking functions take place in a processing channel.

[0080] Steady State: A computational state of an embodiment of a GPS receiver according to the present invention, wherein: (at least) most or (typically) all of the interfering signals (collectively in S) have been identified (i.e., “labeled”); and, (at least) most or (typically) all of the signals (interferers and non-interferers) have achieved “good lock” by the GPS receiver. “Good lock” denotes that the estimates of Doppler, phase and offsets are varying within an acceptable range (e.g., one set of experiments indicated phase must be within 12 degrees of truth, Doppler must be within 28 Hz of truth, and code offset must be within 1/50 of a chip).

[0081] Phase Invariant Process: A phase invariant process requires no knowledge or estimation of the phase of the interfering signal(s), i.e., any phase value produces the same result.

[0082] The present invention is a signal processing method and system for reducing interference so that a receiver can more effectively detect and utilize legitimate wireless signals as well as mitigate, cancel and/or remove interfering wireless signals. The present invention is applicable to a broad range of architectures for processing structured wireless signals in which the interference to be removed can be predictively modeled (e.g., interference whose structure is known and can be simulated). The present invention can therefore be applied to a broad number of structured signal types including radio navigation (including for example GPS, GLONASS and Galileo), RADAR, and cellular signals and in a variety of interference applications, including interfering signals having powers equal and/or unequal to a signal of interest, whether the interference is narrowband, swept, or chirped, and in the removal or isolation of specific signals in a composite signal.

[0083] The present invention is normally not limited to any frequency or modulation scheme. For example, the present invention can be employed with a broad range of frequencies, modulation schemes and PRNs including L1, L2, L5, Galileo's E1 and E2, GLObal NAvigation Satellite System or GLONASS, Code Division Multiple Access or CDMA, Frequency Division Multiple Access or FDMA, Time Division Multiple Access or TDMA, Coarse/Acquisition or C/A code, Precision code or P code, Precision (Encrypted) or P(Y) code, Military code or M code, and Galileo Binary Offset Carrier or BOC codes, and to GPS augmentation methods including Differential GPS or DGPS schemes (which are code- and carrier-based and include both Local-Area and Wide-Area DGPS), Wide Area Augmentation System or WAAS, Local Area Augmentation System or LAAS, European Geostationary Navigation Overlay Service or EGNOS, and Space Based Augmentation System or SBAS.

[0084] In one embodiment, the present invention includes the following steps:

(a) receiving a composite signal (y), y including, from each of a number of signaling sources, a corresponding wireless signal comprising information;

(b) obtaining, for each of the wireless signals, respective signal modeling data indicative of the corresponding wireless signal, the respective signal modeling data representing both in phase and quadrature components of the corresponding signal;

[0085] (c) projecting a representation of y, onto a subspace, the subspace being orthogonal to a space spanning a representation of at least one signal (s₀) of the corresponding wireless signals, the space being determined using the respective signal modeling data for the signal(s) s₀ and the subspace being non-orthogonal to a representation of another signal (h_O) of the wireless signals;
(d) using a result from step (c) to acquire the signal h_O; and
(e) after acquiring the signal h_O, determining the information from the signal h_O.

[0086] The signal processing embodiment is normally performed after the antenna, signal sampling, and analog-to-digital stages in the signal receiver. The invention can therefore work with various antenna architectures and antenna-based interference mitigation techniques, including multiple antennas, multiple-element antennas, adaptive antenna arrays, and multi-beam and adaptive nulling antennas. The embodiment can work with various signal sampling schemes, including direct Radio Frequency or RF sampling, Intermediate Frequency or IF sampling, baseband sampling (e.g., quadrature sampling), and bandpass sampling.

[0087] In DGPS schemes, common (i.e., correlated) errors are preferably removed after the signal processing techniques of the present invention are performed.

[0088] In one configuration, the receiver, when in steady state operation, performs the method of this embodiment by treating all known signals except for the current channel's signal as interference and building the S (interferer) matrix appropriately.

[0089] In another configuration, which is suboptimal, the receiver, during steady state operation, builds the interferer matrix with only the n highest-powered signals. N is the number of interferers and is limited by computational constraints. This configuration can determine the matrix S with a minimal of computational intensity and therefore allow for good performance with lower computational requirements.

[0090] The present invention can have a number of advantages depending on the particular configuration:

[0091] First, signal processing in this method generally does not depend on the particular radio-navigation signal that is transmitted. For example, the present invention can be applied to any signal frequency (e.g. radio frequencies CA2, L1, L2, L5, E1, E2, M1, M2). In addition, it can be applied to any pseudo-random number (PRN) code (e.g., Coarse Acquisition (C/A) code, Precise P(Y) code, Military M-codes, and the to-be-defined E-codes).

[0092] Second, radio-navigation receivers equipped with an embodiment of the invention can be resistant to near-far interference (including special cases such as self-interference, multipath interference, and jamming).

[0093] Third, this invention can eliminate the necessity to determine the phase of the carrier of the interfering signals. Thus, interference mitigation in this invention is a phase-invariant process.

[0094] Fourth, this invention can be used with and transparent to existing navigation augmentation and landing systems (e.g. WAAS, LAAS, and Inertial Navigation Systems (INS)). Furthermore, should these augmentation and landing systems provide ranging information, the present invention can be an integral part of their receiver architectures.

[0095] Fifth, this invention is fully compatible with most current Receiver Autonomous Integrity Monitoring (RAIM) techniques. RAIN provides timely warnings to GPS receiver users when the integrity of their PVT solution has been compromised. The various RAIM techniques are all based on some kind of self-consistency check among the available measurements. To be effective RAIM requires redundancy of information, i.e., 5 satellites to detect an anomaly and 6 satellites to identify and remove its faulty data from the navigation solution. Accordingly, the present invention adds an additional integrity monitoring technique for detecting and preventing smart jamming, and multipath signals.

[0096] Sixth, this invention can make radio-navigation receivers more robust to interference. Embodiments of the invention can operate within a radio-navigation receiver as a signal processing technique. Additionally, embodiments of the present invention can be effectively used on analog or digital signals, and on RF or IF ranges. Thus, if predetermined and/or predictable structured interference is present in a wireless navigation signal, such interference can be removed and the resultant signal is passed to acquisition and tracking routines, as one skilled in the art will understand.

[0097] Seventh, the invention can be embodied in software, firmware or other programmable techniques within a GPS receiver having appropriate hardware to enable the signal processing performed by the present invention. Moreover, substantially all processing performed by the invention is embedded within the logic of one or more special purpose hardware components (e.g., chips, logic circuits, etc.), eliminating the need for programming such hardware components. Of course, hybrid embodiments that are between a substantially programmed embodiment and a substantially hardware embodiment are also within the scope of the present invention.

[0098] Other objects, features, and advantages of the present invention will be apparent from the following detailed description of the preferred embodiment and the accompanying figures herewith.

[0111] Referring to the Wireless Signal Modeling description in the Terms and Definitions section above, Sharf and Friedlander (Sharf L. L., B. Friedlander, “Matched Subspace Detectors,” IEEE Trans Signal Proc SP-42:8, pp. 2146-2157, August 1994 incorporated fully herein by reference), showed that when the measurement noise variance is unknown, the uniformly most powerful (UMP) test for detecting contribution from H, while rejecting contributions from S is given by:

[0112] $\begin{array}{cc}\tau \left(y\right)=\frac{y^T{P}_{G}y}{y^T{P}_S^{\perp }{P}_G^{\perp }{P}_S^{\perp }y}& \left(3\right)\end{array}$
where:

• • [0113] τ(y) is the Generalized Likelihood Ratio Test (GLRT). That is, the test provided by τ(y) is for determining whether the signal H (i.e., the target signal) is part of the composite signal y. More specifically, the signal H is declared present in y if τ(y) exceeds some appropriately defined threshold.
  • [0114] P_S=S(S^TS)⁻¹ S^T is the orthogonal projection operator matrix that would take an input and project it onto the space spanned by the columns of the matrix S.
  • [0115] P_S^⊥=I−P_S is the orthogonal projection operator matrix that would take an input and project it onto the space spanned by the columns of the matrix perpendicular to S.
  • [0116] G=P_S^⊥H=H−P_SH is a matrix whose columns span a subspace that contains those portions of H perpendicular to S. When searching for one signal, the matrix H reduces to a column vector, therefore G reduces to a column vector as well.
  • [0117] P_G=G(G^TG)⁻¹G^T is the orthogonal projection operator matrix that would take an input and project it onto the space spanned by the columns of the matrix G.

[0118] P_G^⊥=I−P_G is the orthogonal projection operator matrix that takes an input and projects it onto the space spanned by the columns of the matrix perpendicular to G.

[0119] Many of the projections are illustrated in FIG. 3. As can be seen from the figure, the subspace 300 corresponds to the detection subspace, and subspace 308 corresponds to the interference subspace and subspace 304 to the signal-of-interest subspace <H>. The interference subspace spans a representation of one or more selected interfering signals and is determined using signal modeling data for the selected interfering signals. The signal “y” refers to a representation of the composite signal received by the antenna, and the subspace 304 is spanned by a representation of the signal of interest (or signal H). As can be seen from FIG. 3, a representation of y is projected onto the detection subspace, which is orthogonal to the interference subspace but non-orthogonal to the signal-of-interest <H> 304 subspace spanned by a representation of the signal of interest, to acquire the signal of interest. The above test using τ(y) has the additional benefit that the performance is invariant to non-negative scalings of y, i.e., multiplyingy by a constant does not impact the test. Most RNSS receivers currently use a Matched Filter Detector (MFD) to perform the following computation:
z=H^Ty  (4)
to determine if the signal H is present in y, i.e., the signal H is declared present in y if z exceeds some appropriately defined threshold. The present invention involves computing an un-normalized version of Equation 3 that has the form of Equation 4, which is suitable for use in today's RNSS receivers with only minor modifications (if any).

[0120] Note that there are normalization terms in both the denominator and numerator in Equation 3. Moreover, the denominator is only a normalization term and is ignored based on the following reasons. In a wireless receiver's tracking stage, (e.g., using FLL, and/or PLL) the ratio of the in-phase and quadrature signal components in the discriminator cancel any scale factor that results from the lack of normalization (e.g., such components included in component 428 of FIG. 4 described herein below). Additionally, in acquisition, computing a properly adjusted detection threshold will also account for the lack of normalization.

[0121] The numerator term y^TP_Gy of Equation 3 can be decomposed as follows:
y^TP_Gy=y^TG(G^TG)⁻¹G^Ty  (5)
y^TP_Gy=y^TG(G^TG)^−1/2(G^TG)^−1/2G^Ty  (6)
y^TP_Gy=((G^TG)^−1/2G^Ty)^T(G^TG)^−1/2G^Ty  (7)

[0122] The final form of the numerator term in the above Equation 7 implies that a segment of y is operated on by the term (G^TG)^−1/2GT. The result is then squared to compute y^TP_Gy. Clearly, the term that is applied to the segment y is normalized, where the normalization term is (G^TG)^−1/2. To reiterate, the present invention modifies the computations provided in the Sharf and Friedlander, 1994 reference cited above, and performs an un-normalized operation on the measurement segment, y, by eliminating the term (G^TG)^−1/2 for the same reasoning that allows the denominator of Equation 3 to be ignored.

[0123] Accordingly, the detection test can now be written in a form similar to Equation 4. The resulting operation on y is given by:
z=G^Ty  (7a)
Thus, the signal H is declared present in y if G^Ty exceeds some appropriately defined threshold.

[0124] Recall that G=P_S^⊥H and therefore z from the above equation (7a) can be re-written:
z=H^TP_S^⊥y  (8)
By defining {tilde over (y)} as:
{tilde over (y)}=P_S^⊥y  (9)
another form of a detection test similar to the conventional receiver test of Equation 4 is obtained as follows:
z=H^T{tilde over (y)}  (10)

[0125] Note that as per Equation 9, {tilde over (y)} is the projection of y onto the perpendicular space spanned by the columns of S, which is oblique (i.e., non-orthogonal) to H. In other words, once {tilde over (y)} has been calculated, the signal with interference S removed, it is simply passed to the standard acquisition and tracking stages.

[0126] To perform a phase invariant MSD, the signal, y, interference, S, and target signal, H, are generally complex, i.e., their real and imaginary parts are in-phase-quadrature couples. Given this, the equations above are also valid for the phase-invariant MSD with the understanding that the T operator which represents a transpose above must be replaced with an adjoint operator, such as a Hermitian transpose.

[0127] The phase invariant MSD incorporates phase dependence in interference mitigation by using both the in-phase and quadrature components to represent selected signals. The signal measured at the antenna (y) is represented using in-phase and quadrature (I & Q) signal components. The modeled interference (S) is represented using both I & Q signal components. Finally, the modeled reference signal (H) (the signal of interest), is represented using both I & Q signal components. As before, S and H can be a vector or a matrix of vectors.

[0128] When I & Q components are employed to represent a signal, the signal's vectors become vectors of complex numbers. For example, a target signal h₁ is composed of a real component (hI₁, a column vector containing the in-phase component) and a complex component (i*hQ₁, where i=sqrt(−1) and hQ₁=a column vector containing the quadrature component) as follows:
h₁=hI₁+i*hQ₁  (11)

[0129] The construction of H and S matrices then follows as described above, except that complex vectors may be used in place of real vectors.

[0130] The phase invariant signal modeling technique is performed after the antenna, and analog-to-digital conversion (ADC) stages in structured signal receivers. The technique can work with various antenna architectures and antenna-based interference mitigation techniques. Such architectures and interference mitigation techniques include multiple antennas, adaptive array, multibeam, and adaptive nulling antennas. The technique can work with various sampling schemes including but not limited to direct RF sampling, IF sampling, baseband sampling (a.k.a. quadrature sampling), and bandpass sampling. The technique can also work with notch filters employed for suppression of narrow-band structured interference.

[0131] In contrast, the GPS receiver of U.S. Pat. No. 6,771,214, which is incorporated herein by this reference, discloses interference mitigation techniques requiring explicit knowledge of code offset, Doppler, and phase to model the interference (S) and the reference signal (h). Code offset is usually provided by the Delay Locked Loop or DLL and phase from the Phase Locked Loop or PLL. The change in Doppler when using a PLL is usually obtained by integrating the phase error. The present invention presents a method where phase for the interfering signal(s) can be incorporated implicitly and therefore avoids the use of a tracking loop. The methodology of the present invention can be invariant not only to interference power but also to interference phase.

[0132] The phase invariant signal modeling technique can be applied to hardware (e.g., using a logic circuit such as an Application Specific Integrated Circuit), software, and/or firmware receivers. The technique can be implemented in serial, parallel, and/or massively parallel architectures and using both conventional and block processing methods.

[0133] Because the phase invariant signal modeling technique requires only that structured interference can be modeled, the technique is applicable to a wide range of problems involving signal detection in the presence of structured interference. These problems include, for example, near-far interference or interference arising from signals having unequal powers, interference signals having powers equal to the signals of interest, and the removal or isolation of specific signals in a composite signal.

[0134] The phase invariant signal modeling technique can have a number of other applications. For example, the technique can be used in stationary, dynamic, highly dynamic, ground-based, marine-based, air-based or space-based sensing platforms, whether in monostatic and bi-static remote sensing applications; in co-site interference applications; and, in radio navigation applications. Furthermore, this technique when implemented in a GPS receiver may be integrated with other navigation sensors, including inertial sensors, dopplometers, altimeters, speedometers, and odometers. The technique could enable a signal processing platform to have both a transmitter and receiver operating simultaneously without experiencing unacceptable levels of emitted signal-related noise in the received signal. A remote sensing platform so equipped can eliminate eclipsing in the returns.

[0135] An alternate variant to the solution of {tilde over (y)} presented in U.S. Pat. No. 6,771,214 is to process y and the S matrix using the Gram-Schmidt process. This process, which is particularly useful for radar and GPS applications, is done iteratively on vector components and thus eliminates the matrix inverse which can be complex to implement in hardware.

[0136] This solution requires that the columns of S are normalized, i.e., unit vectors. The following MATLAB code will be used to illustrate this solution:

[0137]

	%-----------loop over the number of columns of S
	[line 1]	for i=1:numColumns
	[line 2]	[val,index]=max(y′*S);
	[line 3]	Sbig=S(:,index);
	[line 4]	S(:,index) = [ ];
	[line 5]	y=y−(Sbiq′*y)*Sbig;
	[line 6]	[rowb,colb]=size(S);
	[line 7]	Snew=[ ];
	[line 8]	for j=1:colb
	[line 9]	Snew=[Snew (S(:,j) − (Sbig′*S(:,j))*Sbig)];
	%-------------------renormalize S
	renormalize S
	[line 10]	S(:,j)=Snew(:,j)/norm(Snew(:,j));
	[line 11]	end;
	[line 12]	end;

line 1 will perform the following process a number of times equal to the columns of matrix S, where i is the column identifier.
line 2 is the maximum inner product value of a column of S with y

[0138] $\begin{array}{cc}\underset{j}{\max }\left\{y^T{S}_j❘S_j\in S\right\}& \left(12\right)\end{array}$
line 3 is the column of S that corresponds to the max, defined to be {right arrow over (s)}_j (or Sbig)
line 4 eliminates that column, {right arrow over (S)}_j, from the S matrix
line 5 calculates the portion of {right arrow over (y)} that is orthogonal to
y=y−({right arrow over (s)}_j^Ty){right arrow over (s)}_j  (13)
line 6-11 performs the same calculation on each column of the remaining S matrix
S=[{right arrow over (s)}₁{right arrow over (s)}₂ . . . {right arrow over (s)}_n]  (13)
{right arrow over (k)}_k={right arrow over (s)}_k−({right arrow over (s)}_j^T{right arrow over (s)}_k){right arrow over (s)}_j  (14)
As will be appreciated, “rowb” and “colb” refer, respectively, to the number of rows and columns remaining in the S matrix after removal of Sbig. Snew refers to the updated matrix variable.
In line 8, j is a column counter that progressively causes processing of each of the remaining columns, colb, in the S matrix.
line 10 ensures that the columns of S are still unit vectors
{tilde over (y)} is equal toy after the algorithm is complete.
As can be seen from the two end commands, the code creates two loops, the first is created by line 1 and the second by line 8. The second loop is nested in the first loop.
The preceding MATLAB code should not be taken to be the preferred or optimal instantiation of this technique, for it has been written for clarity. Other instantiations can eliminate the need to normalize some of the vectors. As mentioned earlier this technique eliminates the need to calculate the inverse and is therefore immune to the potential numerical instabilities associated with an inverse. This technique, falls into the optimal class as does the preferred implementation mentioned in patent in Data Fusion Corporation's U.S. Pat. No. 6,771,214, “GPS Near-Far Resistant Receiver”. U.S. Pat. No. 6,771,214 dealt exclusively with GPS and GNSS applications. This patent covers the unique applications and modifications needed for additional signal processing domains, including the radar and remote sensing domains.

[0139] In spread spectrum techniques like CDMA, each user is assigned a time varying code that is used to spread each bit in the digital data stream to occupy the entire spectral band allocated to the Multiple Access System (MAS). The different users in such a system are distinguished by the unique spreading codes assigned to each user. In this way, all users simultaneously employ all of the bandwidth during wireless communications.

[0140] Each user in a CDMA system uses a unique noise (pseudorandom PN) code to spread the bits. Upon reception, each user's coded signal is compressed using a matched filter that is matched to that user's code to extract the desired bit sequence.

[0141] Ideally the user codes are designed to be perfectly orthogonal when the codes are aligned. Such an alignment of the codes is achieved when there is perfect synchronization. In such a case, the matched filtering operation—which is based on an orthogonal projection—completely nulls out all users except the user of interest. This is illustrated in FIG. 1, where the user code of interest is the vector H that is shown to be orthogonal to S, the matrix of all the other user codes. As shown in this figure, the orthogonal projections of y (a measurement vector of the composite signal) onto H using a matched filter results in perfect cancellation of all contributions from S.

[0142] However, GPS codes are not perfectly orthogonal. In particular, the C/A-code length is only 1,023 chips. So the cross-correlation properties can be poor under certain circumstances (as discussed in Parkinson, B. W., Spilker, J. J., Global Positioning System Theory & Applications, vol-1&2, American Institute of Aeronautics and Astronautics, 1996, incorporated herein by reference). Alternatively, the P(Y) code with 6.1871×10¹² chips is virtually orthogonal for all offsets. However, the problem with P(Y) code is that reception of the full code literally takes a week and the required integration time is computationally staggering. Accordingly, all P(Y) code receivers use significantly shorter code lengths, which again compromises its cross correlation properties.

[0143] The non-orthogonality of comingled signals is further exacerbated by asynchronous communications among the receiver and satellites. Therefore the GPS codes will never be perfectly aligned in time. Thus, if a matched filter is used to decode a desired GPS signal from a satellite or PL, signals from other satellite or PL signals will leak into the desired signal channel as shown in FIG. 2.

[0144] The signal leakage from other satellites or PLs becomes interference noise against which the desired target signal must be detected. As the interference noise-level increases, the performances of the receiver's detectors degrade. As discussed in the Terms and Definitions section, this effect is commonly referred to as the “near-far” effect.

[0145] The present invention may be implemented as a signal processing technique in a radio-navigation receiver. Radio-navigation receivers vary widely in capabilities and design but embodiments of the present invention can be included therein in much the same way regardless of the wireless navigation receiver design. For discussion purposes, FIG. 4 is indicative of many of today's radio-navigation receivers.

[0146] FIG. 4 is a diagram of a generic GPS receiver 400, with the invention included. Note that specific receiver design is usually proprietary. Thus, variations on the diagram of FIG. 4 can include: (i) a GPS receiver that is entirely analog, (ii) a GPS receiver wherein processing is performed in the RF range, and (iii) a different placement of one or more of the pre amp 404, the A/D converter 412, etc. A high level description of the components of the GPS receiver 400 and their signal processing relationship to one another will now be provided.

[0147] The GPS radio-frequency signals of all space vehicles (SVs) in view are received by the antenna 401. Typically the antenna 401 is right-hand circularly polarized with nearly hemispherical gain coverage. The RF signals are amplified by a low-noise preamplifier 404. This amplifier is by far the major contributor of noise to the receiver 400. In some implementations a passive bandpass filter is located between the antenna 401 and the preamp 404 to remove out-of-band RF interference. The amplified and signal-conditioned RF signals are then down converted, via the down converter 408, using signal mixing frequencies from local oscillators 420 (LOs). For the present invention, the down conversion must produce a complex signal by demodulating with both a cosine and a sine term. The input frequencies to the LOs 420 are derived from the reference oscillator 448 via the frequency synthesizer 444 based on the frequency plan of the particular receiver design. There may be one LO 420 per down converter 408. The upper sidebands and leak-through signals are passed through a postmixer bandpass filter (not shown) of the down converter 408 to complete the down conversion process. The analog to digital converter (ADC) operates at IF or baseband. The digital signal is then split once for each channel, and the signal for each channel is passed through a MSD 492 to have the interference removed. From here the channel signals are passed to the normal N digital receiver channels 424. Using the output from MSD 492 each receiver channel 424 performs the acquisition functions and the code and carrier tracking loops (i.e., delay locked loop (DLL), frequency locked loop (FLL), and phase locked loop (PLL)) for a single SV or PL. Or, depending on the receiver, the digital receiver channels primarily perform the correlation operations and work with the receiver processing component 428 for implementing loop discriminators and filters, data demodulation, meters, phase lock loops, and so forth. Additionally, the receiver processing component 428 may store various types of data in, e.g., a data store 426 for each channel 424 which is tracking a signal. This data store is shared amongst all the channels 424. After the data obtained from the GPS signals has been demodulated it is passed to the navigation processing component 436 and the position, velocity and time (PVT) solution is calculated, and displayed 440.

[0148] Within a receiver 400, an embodiment of the present invention may be incorporated into the receiver channel(s) 424 and the receiver processing component 428. For a digital implementation, however, the present invention requires that the dynamic range of the ADC 412 be sufficient to capture all the signals of interest, including the interference. Similarly components 424 and 428 may need to be modified or replaced depending on the dynamic range of the ADC 412 and the requirements of the invention (e.g., the data store 426 and additional processing capabilities).

[0149] The processing channels 432 (FIG. 4) identifies the components within which the present invention is, in at least one embodiment, incorporated. For each interference mitigation performed by the present invention, the data store 426 stores the following parameters: (a) a pseudo-random number, (b) code offset data obtained from the delayed lock loops (DLL), or Kalman filter, (c) Doppler data obtained from the frequency locked loop (FLL) or Kalman filter, and (d) at least one bit that can be toggled between being set and being unset depending on whether the its corresponding GPS signal has been labeled an interferer or not. As noted, the data store does not store phase data obtained from the PLL. This is different from the receiver of U.S. Pat. No. 6,771,214.

[0150] When near-far interference effects occur, the above identified parameters (a) through (d) are usually sufficient to model the interference of each dominant interfering GPS signal. In a more demanding situation (e.g., relatively lower power interferers), a Kalman filter or higher order DLLs, FLLs and PLLs (as described in the Terms and Definitions section above) may also be used to provide code offset rate, and Doppler rate to model the interference signals more accurately when provided with the output of the A/D converter 412. Note that it is within the scope of the present invention to obtain such parameters by other techniques as well, e.g., a massively parallel acquisition scheme (as described in the Terms and Definitions) may provide the required modeling parameters at the fidelity necessary.

[0151] The invention predictively models identified interfering signals. In GPS NF situations, the interference includes one or more SV's (and/or PL's) pseudo-random noise code (PRN). Such interference can be predictively modeled, since all GPS signals can be predictively modeled (in that the structure of such signals is well known and accordingly can be substantially reconstructed from such signal modeling parameters as identified hereinabove). In some situations, such signals may be known a-priori, in which case the standard baseline receiver 400 architecture of FIG. 4 can acquire and track such interferers normally (i.e., as any other GPS signal), and then the procedure of FIG. 7 (described herein below) can be used to acquire and track each desired nominal (i.e., non-interference) signal by canceling out the a-priori interferers. If the interfering PRN signals are not known a-priori, the procedures of FIG. 5 and FIG. 6 (also described herein below) have been demonstrated to cancel the interferers in GPS.

[0152] In extreme cases of NF interference, the signal power levels (from distinct GPS signaling sources) are dramatically different from each other, (e.g., one PL's signal, call it PL1, might be 20 dBW above the nominal satellites' signals (i.e., GPS constellation satellites operating normally) and another PL's signal, call it PL2, might be 20 dBW above that of PL1). In this situation, the interferers must be iteratively identified from highest power to lowest power as FIG. 5 illustrates. Note that the procedure of FIG. 5, in general, takes more elapsed time to accomplish than the procedure illustrated in FIG. 6 but tends to be more robust.

[0153] A description of the steps of FIG. 5 follows. Note that these steps refer to the components of FIG. 4; however, it is to be understood that the A/D converter 412 and the components of the interference processing channel 432 must include the requisite functionality to perform these steps (as well as the steps of FIG. 7),

[0154]

Step	The receiver 400 attempts to acquire and track all observable GPS
504:	signals. This is primarily done using a standard GPS acquisition
	and tracking functions components 424 & 428.
Step	Once the signals that can be tracked have achieved ‘good’ lock,
508:	(i.e., the estimate of Doppler offsets is varying within an
	acceptable range), the most powerful signal is compared to a
	threshold criterion or threshold criteria (without loss of generality,
	henceforth referred to in the singular). This threshold criterion may
	be as simple as identifying the strongest M of N signals as
	presumed interferers, or the threshold criterion could be based on
	design specifications such as those of USCG“, Navstar GPS Space
	Segment/Navigation User Interfaces”, ICD-GPS-200C
	incorporated herein by reference. (E.g., the threshold could be a
	predetermined power level value, such as 10 dBW, above the user
	minimum received power using a 3-dB gain linearly polarized
	antenna. The specified user minimum received power for L1 C/A-
	Code is −160 dBW, for L1 P(Y)-Code is −163 dBW and for L2
	P(Y)-Code or C/A-Code is −166 dBW). This threshold criterion,
	in some embodiments, may be defined as a value corresponding to
	a signal to noise ratio (SNR), or signal to carrier ratio (SCR), or
	carrier to noise ratio (CNR), relative amplitude, or some
	combination of such values (e.g., a weighted sum thereof or a
	series of threshold tests wherein each test uses at least one of these
	ratios). Note that the Parkinson, B. W., Spilker, J. J., 1996
	reference cited hereinabove (and incorporated fully herein by
	reference) details such a method for combining such values.
Step	Any signal exceeding (or violating) this threshold criterion is
512:	labeled as an interferer (i.e., the data representing such a signal is
	accordingly labeled such as by toggling the bit field (d) of the
	modeling parameters described hereinabove). This labeled
	interferer continues to be tracked and it's current Doppler and code
	offset are saved in the data store 426.
Step	All signals that have not yet labeled as interferers are then re-
516:	acquired and tracked by invoking FIG. 7 to cancel out all labeled
	interferers to obtain better acquisition and tracking of the desired
	signals. Control then passes back to step 508.
Step	After all the interfering signals have been labeled, and all the non-
520:	interfering signals have achieved good lock, the receiver enters
	what is denoted wherein as a ‘steady state’. Tracking continues for
	all channels by, again, performing the procedure described in FIG.
	7, and the navigation data is demodulated.
Step	Once sufficient navigation data has been demodulated, the desired
524:	navigation solution (e.g., PVT) can be computed via the navigation
	processing component 436.

[0155] FIG. 6 is similar to FIG. 5 except that the identification (i.e., label) of all initial interferers is done at one time. (FIG. 5 demonstrated an iterative scheme that identified interferers from highest power downwards). This procedure has been shown to be useful in cases when all the interfering GPS signals have roughly of equal power.

[0156]

Step	The receiver attempts to acquire and track all observable GPS
604:	signals. This is primarily done using the standard GPS acquisition
	and tracking functions components 424 and 428.
Step	Once the signals that can be tracked have achieved ‘good’ lock,
608:	(i.e., the estimate of Doppler offsets is varying within an
	acceptable range, the most powerful signal is compared to a
	threshold criterion or threshold criteria (without loss of generality,
	henceforth referred to in the singular). This threshold criterion
	may be as simple as identifying the strongest M of N signals as
	presumed interferers, or it could be based on design specifications
	such as those of ICD-GPS-200C identified hereinabove (e.g., 10
	dBW above the user minimum received power using a 3-dB gain
	linearly polarized antenna. The specified user minimum received
	power for L1 C/A-Code is −160 dBW, L1 P(Y)-Code is −163 dBW
	and L2 P(Y)-Code or C/A-Code is −166 dBW). This threshold
	criterion, in some embodiments, may be defined as a value
	corresponding to a signal to noise ratio (SNR), or signal to carrier
	ratio (SCR), or carrier to noise ratio (CNR), relative amplitude, or
	some combination of such values (as in step 508 above).
Step	Any signal exceeding this threshold is labeled as an interferer.
612:	These labeled interferers continue to be tracked and their current
	Doppler and code offset are saved in the data store 426.
Step	All signals that are not labeled as interferers are then re-acquired
616:	and tracked by invoking the steps summarized in FIG. 7 to cancel
	out all labeled interferers and thereby to obtain better acquisition
	and tracking of the desired signals. The navigation data is
	demodulated.
Step	Once sufficient navigation data has be demodulated the navigation
620:	solution can be computed 436.

[0157] FIG. 7 shows a high level process diagram of the operation of the signal interference mitigation method and apparatus of the present invention, during steady state. Steady state is defined in the Terms and Definition section above. The steps of FIG. 7 may be described for a GPS embodiment as follows:

[0158]

Step	Digital data y (as described in Interference Mitigation Technique
704:	section above) is provided to the interference processing
	channel432. As mentioned earlier this data could be IF or RF, but
	it must be complex with the imaginary and real parts being in-
	phase quadrature pairs. In FIG. 4 the data y is received by the
	antenna 401 and then passed to the pre-amp 404 for amplification.
	Assuming the GPS radio-navigation signal y is then down
	converted, via down converter 408, to an intermediate frequency
	(IF) and passed through to the A/D converter 412 for digitally
	sampling. The number of bits in the A/D converter 412 must be
	sufficient to capture all signals of interest, including the
	interference signals. Accordingly, for an embodiment of the
	receiver 400 to capture, e.g., 7 to 8 signals differing by +33 dB, the
	number of bits in a linear A/D converter 412 is at least 8 bits. Note
	that when near-far interference effects occur, the power of at least
	one interfering signal is significantly higher than any non-
	interfering (nominal) GPS signals, thus requiring the A/D
	converter 412 to utilize such a range of bits to be able to acquire
	and track the weaker signals. It is the relative power difference
	that is important, (e.g., nominal signals could be the NF
	interference to extremely weak signal as well). A GPS receiver
	needs at least 1 bit to acquire and track a GPS signal in thermal
	noise. After passing through the ADC 412, the IF signal is fed to
	all processing channels.
Step	Retrieve the current interference modeling parameters from the
708:	data store 426 having interference modeling parameters for all the
	interference processing channels 432. Note, there may be no
	current interferers, or there may be one or multiple interferers. All
	the interference modeling parameters (for signals that have labeled
	as interferers) that exist in the data store 426 are retrieved.
Step	If any interference modeling parameters are retrieved in step 708,
716:	then control is passed to 720, else control goes to 728.
Step	The interference modeling parameters are used to create an in-
720:	phase and quadrature complex interference reference signals which
	have the form:
	siCA (ωt) = [Gi(t)]eiωt	(15)
	where:
	Gi(t) is the Coarse Acquisition (C/A) code's ith chip at time t (used
	in the Standard Positioning System (SPS))
	ω is the carrier frequency.
	Once the interference signal of each interference PRN is
	created, the complex interference signals are assembled into a
	matrix, S, as in equation (2) of the Wireless Signal Model
	description in the Terms and Definitions section hereinabove.
	Note, the desired signal (for which interference is being removed)
	will not be part of S even if it is identified as an interferer. This
	rule even applies to multipath, and other jamming signals.
Step	Once S has been calculated, {tilde over (y)} is determined as per equation (9),
724:	thereby removing the interference characterized by the interference
	modeling parameters.
Step	Perform conventional acquisition and tracking functions. If this
728:	step was entered from step 724 use {tilde over (y)} (for computing Equation 10)
	and if this step was entered from 716 use y (for computing
	Equation 4).
Step	The desired signal is compared to the interference threshold
734:	criterion that was used in 508 or 608. If the classification of the
	desired signal is still labeled as an interferer (based on info from
	the data store 426), or if it has changed from being an interferer to
	a non-interferer control will pass to step 738, otherwise control of
	the interference processing channel 432 passes to the next step of
	FIG. 5 (or FIG. 6) following the current activation of FIG. 7.
	Applicants have observed that a potential interfering signal
	should exceed (or violate) the threshold criterion for at least twenty
	times in a row before its classification should be changed and
	therefore labeled an interferer (or unlabeled thereby designating
	that it is not an interferer). This prevents the signal's classification
	from being changed mistakenly due to a noise spike or a navigation
	bit flip.
Step	Update the desired signal's (this channel's) current interference
738:	modeling parameters to the data store 426 of all the interference
	processing channels. Note such updates include: (i) removing a
	labeling of the desired signal as an interferer if this signal remains
	below the threshold criterion consistently, and (ii) labeling the
	desired signal as an interferer if this signal was consistently above
	the threshold criterion or just updating the current signals
	interference modeling parameters.

[0159] FIG. 8 illustrates a single receiver channel. A GPS receiver commonly has many such channels, e.g., all in sky receivers typically have at least 12 channels. Both the inphase and quadrature components of the signal from the Receiver Frontend 846 is split and passed to each channel's MSD 812. The Doppler and chip-offset parameters for each interfering PRN are passed to the MSD 812 from the universal Data Store 828. Once the interference has been removed, the signal is passed to a baseline receiver channel 824. Each channel performs a matched filter for a single PRN. Traditionally, first C/A code acquisition is performed. Then tracking is performed using the Doppler and chip-offset measurements from the acquisition step. The parameters from the tracking step are passed to the Data Store 828 and used in P(Y) code acquisition. Finally, P(Y) tracking is performed, and the resulting output is stored in the Data Store 828. At any time, the Navigation Processor 836 can use the data in the Data Store to determine the receiver's position, velocity, and time. FIG. 8 can also be used for other GPS signals and codes. For example, the receiver could directly acquire the P(Y) code, or M-code, etc.

[0160] The present invention is directly applicable to all known GNSS codes. For example, applying this invention to the Precise Positioning System (PPS) as well as the Standard Positioning System (SPS) using the CA code discussed above. The GPS signal structure for the L1 frequency looks like:
L1_i(ωt)=A[P_i(t)⊕D_i(t)] cos(ωt)+√{square root over (2)}A[G_i(t)⊕D_i(t)] sin(ωt)  (16)

[0161] where:

[0162] A is the amplitude

[0163] P_i(t) is the Precise [P(Y)] code's i^th chip at time t [used in PPS]

[0164] D_i(t) is the nav data's i^th bit at time t

[0165] G_i(t) is the Coarse Acquisition [C/A] code's i^th chip at time t [used in SPS]

[0166] ω is the carrier frequency

[0167] As can be seen, the C/A code is 3 dB stronger than the P(Y) code. If a receiver 400 is sampling the signal at a rate sufficient for CA code, then the P(Y) will appear as noise 3 dB less than the original interference signal. Long coherent integration can mitigate the effect of the strong P(Y) code noise.

[0168] If a PPS GPS receiver is being used, the sampling rate will be 10 times higher than the sampling rate needed for the CA code. Most GPS receivers acquire and track the CA code and use the information in the navigation message to ‘jump’ into the P(Y) code. It is also possible to directly acquire the P(Y) code, as one of ordinary skill in the GPS field understands.

[0169] Because of the higher sampling rates the noise/interference from the P(Y) code can now be removed. The interference associated with the P(Y) code can be removed in exactly the same way as was done for the CA code. This process is illustrated in the flowchart of FIG. 9 whose steps will now be described:

[0170]

Step 904:	Digital data y (as described in Interference Mitigation
	Technique section above) is provided to the interference
	processing channel 432. As mentioned earlier this data could
	be baseband, IF or RF. There are numerous possible designs
	for a RF front end, the important point is that this invention
	obtains complex digital data from the front end. This digital
	data is then split and passed to the receivers various channels.
Step 908:	Retrieve the current interference modeling parameters from
	the data store 426 having interference modeling parameters
	for all the interference processing channels 432. Note, there
	may be no current interferers, or there may be one or multiple
	interferers. All the interference modeling parameters (for
	signals that have labeled as interferers) that exist in the data
	store 426 are retrieved.
Step 916:	If any interference modeling parameters are retrieved in step
	908, then control is passed to 920, else control will go to 928.
Step 920:	Once the P(Y) code and the CA code tracking loops are both
	locked onto the signal, the interference to both components
	can be removed at the same time. The interference modeling
	parameters are used to create an in-phase and quadrature
	complex reference signals which have the form:
	s i ⁡ ( ω ⁢ ⁢ t ) = P i ⁡ ( t ) ⁢ ⅇ ⅈω ⁢ ⁢ t + G i ⁡ ( t ) ⁢ ⅇ ⅈ ⁡ ( ω ⁢ ⁢ t ⁢ π 2 ) ( 17 )
	where the variables have the same meanings as described in
	step 720.
	Once the interference signal of each interference PRN is
	created they are assembles into a matrix, S, as in equation (2)
	(of the Wireless Signal Model description in the Terms and
	Definitions section hereinabove). Note, the desired signal will
	not be part of S even if it is identified as an interferer. This
	rule even applies to multipath and multipath-like signals.
Step 924:	Once S has been calculated, {tilde over (y)} is determined as per
	equation (9), thereby removing the interference characterized
	by the interference modeling parameters.
Step 928:	Perform conventional acquisition and tracking functions. If
	this step was entered from step 924 use {tilde over (y)} and if
	this step was entered from 916 use y.
Step 934:	The signal is once again compared to the interference
	threshold. If the signals classification is still labeled as an
	interferer or it has changed from being an interferer to a
	non-interferer control will pass to step 938 else the control
	interference processing channel and baseline receiver function
	will continue. As mentioned above, step 734, applicants have
	observed that a potential interfering signal should exceed (or
	violate) the threshold criterion for at least twenty times in a
	row before its classification should be changed and therefore
	labeled an interferer (or unlabeled thereby designating that it
	is not an interferer). This prevents a signal's classification
	form changing mistakenly due to a noise spike or a navigation
	bit flip.
Step 938:	Update the desired signal's (this channel's) current
	interference modeling parameters to the data store 426 of all
	the interference processing channels. Note such updates
	include: (i) removing a labeling of the desired signal as an
	interferer if this signal remains below the threshold criterion
	consistently, (ii) labeling the desired signal as an interferer
	if this signal was consistently above the threshold criterion or
	(iii) just updating the current signals interference modeling
	parameters.

[0171] Optimum performance of the present invention is achieved by treating all known signals except for the current channel's signal as interference, and building the S matrix appropriately. A simple suboptimal embodiment of the invention may be created that allows for good performance with lower computational requirements. This is accomplished by building the S matrix with only the n highest-powered signals (not including the current channel's signal) where n is the number of interferers and is limited by computational constraints.

[0172] In other words, for a selected signal, the corresponding identified group of interferers contains only the signals having powers exceeding, by a selected threshold (which may be zero), the power of the selected signal. Interference from signals having powers weaker than the selected signal generally is not a significant obstacle to tracking and acquiring the selected signal. Because the set of interferers contains fewer members relative to the other embodiments of the invention, the embodiment, though being suboptimal, can provide a substantial savings in computational requirements.

[0173] Since the number of channels and computational power are limited, control logic was designed to remove interference efficiently when operating in steady state mode. This method was designed to minimize computational intensity without harming performance. This technique based on the relative post correlation observed powers of the various signals. The procedure for the Interference Subspace Discriminator 1004 is shown in FIG. 10. Each channel will be processed in parallel, and the steps for the i^th channel are detailed below:

[0174]

Step	Loop over all visible PRNs, j ≠ i, where i and j are channel
1006:	identifiers, with each channel being associated with a different
	signal.
Step	Retrieve parameters for the ith and jth PRN from the Data Store
1008:	1028 for the previous time-step. If the power of the jth channel
	is greater than the power of the ith channel to within some
	threshold (to be determined by the specific situation), then
	the jth PRN is treated as an interferer and added to the inter-
	ference matrix for the ith PRN 1012. Otherwise, it is ignored.
	This step is repeated for all visible PRNs, j ≠ i.
Step	The newly created interference matrix, S, from the Interference
1092:	Subspace Discriminator 1004 is used with the signal from the
	Receiver Frontend 1046 and the interference it mitigated in
	the MSD 1092.
Step	The code offset and Doppler of the ith signal is tracked, and the
1056:	output parameters (power, Doppler, chip-offset, phase, and
	navigation data) are sent to the Data Store 1028.
Step	The time step is incremented.
1058:
Step	The interference matrix for the ith signal is deleted, and control is
1060:	returned to the Interference Subspace Discriminator 1004.

[0175] Before the receiver can operate in steady state mode, it must first initialize and populate the Data Store with most of the visible PRNs. FIGS. 5& 6 can be used to populate the Data Store. In another embodiment of the invention the following logic can be also be used to build the Data Store. In general, the logic, after initial signal tracking and acquisition, identifies the “hottest” signals, or the signals having the highest powers, and groups the signals in a first signal grouping. Using the nonorthogonal projection algorithm above, the logic then removes, from the composite signal, the interference attributable to the grouped signals, and tracks and reacquires the ungrouped signals. The “hottest” of the tracked and acquired signals (or the signals having the highest powers among the ungrouped signals) are then grouped in a second signal grouping (having a different membership than the first signal grouping). The interference from the members of both the first and second signal groupings are removed from the measured composite signal and the process repeated until no ungrouped signals remain. Typically, the powers of the signals within any signal grouping fall within a specified or predetermined signal power range, with the various groups having no overlapping signal ranges. This process is shown in more detail in FIG. 11. Referring to FIG. 11:

[0176]

Step	Attempt to acquire and track all known signals in the receiver's
1104:	radio spectrum. Alternatively one could just attempt to acquire and
	track known signals visible to the receiver.
Step	Have all the signals been grouped based upon their relative
1108:	powers? These groupings will be used to guide the interference
	mitigation.
Step	Find the non-grouped signal with the maximum post correlation
1112:	power. Begin group i and add modeling parameters to the group.
Step	Loop over all the remaining non-grouped signals.
1116:
Step	Have we finished looping over all the non-grouped signals?
1120:
Step	Determine if the signal j is in the ith power grouping.
1124:
Step	Add this signal's modeling parameters to the ith power group.
1128:
Step	Re-acquire and track all non-grouped signals by performing the
1132:	steps of FIG. 7 to Cancel out the interferers and obtain better
	acquisition and tracking. This will be done using the known inter-
	ference power groups. Specifically, interference to a particular
	signal is all the power groups having a higher power than the group
	that the particular signal is in.
Step	The receiver is now in steady state. The receiver will transition to
1136:	steady state power control logic.

[0177] If GPS receivers are in a jamming interference environment the present invention provides interference resistance that can be used to eliminate jamming from structured interference. The jamming signal is simply acquired, tracked, and then eliminated from the composite signal by the present invention.

[0178] Self interference is when a GPS transmitter and receiver are substantially colocated. The application of the present invention to mitigate self-interference is actually more accurate than in the standard interference resistant GPS receiver. The reason for this is rather than estimating the interfering GPS signal's offset and Doppler from the tracking loops, this data is supplied exactly by the GPS receiver enhanced according to the present invention.

[0179] Precision navigation and landing systems require reliable and highly accurate position, velocity and time (PVT) information not achievable by standalone GPS. To meet these requirements additional GPS transmitters are needed to improve the accuracy. These additional transmitters can be additional satellites as specified in the Wide Area Augmentation System (WAAS) or PLs based on the ground as specified in the Local Area Augmentation System (LAAS), or on board ships, or even Unmanned Aerial Vehicles (UAV) loitering in the air above an area of interest. WAAS and LAAS can transmit either GPS correction data (i.e., GPS differential data) or provide additional ranging information. For such applications, PLs can be used for Carrier-Phase Differential GPS navigation, resulting in a potential range precision of about 1 mm (as recited in Elrod, B. D., Van Dierendonck, A. J., “Pseudolites”, Ch 2. pg 51 in Global Positioning System: Theory and Applications, Volume II, Ed by B. Parkinson, et al, 1996). When these transmitters use the GPS spectrum, as is the case for UAVs, PLs, and satellites providing ranging information, additional constructive interference is added. In the case of landing systems, a GPS receiver that is too near any one of the PLs will suffer interference. This effectively “drowns out” the reception from the other PLs as well the other GPS satellite signals. When this happens, the GPS receiver will be unable to track the satellites and therefore will be unable to provide PVT information. This is the Near-Far effect. Numerous solutions have been proposed and investigated to mitigate the PL interference, including pulsing the PLs. None of these solutions are as effective as the present invention.

[0180] In particular, the present invention is different from the “pulsing” PL solution described in the Elrod & Dierendonck reference cited above. The Pulsed PL technique with 10% duty cycle (transmitting 93 out of 1023 chips each cycle of the PL signal) may result in a 10 dB improvement in signal to interference level. This amount of improvement was suggested in Stansell, T. A., “Recommended Pseudolite Signal Specification,” in Global Positioning System, vol-3, Institute of Navigation, 1994 also fully incorporated herein by reference. Such a 10 dB improvement is not sufficient in radio environments where PL signals are >30 dB higher than nominal GPS satellite signals. Moreover, if multiple PLs are used, then it becomes harder to maintain time slots for each of the PLs and also the oscillators of the PLs have to be very stable. This is discussed in Parkinson, B. W., Spilker, J. J., 1996 referenced above. It is important to note, however, that the present invention can also be used in conjunction with such pulsed pseudolite technology.

[0181] The present invention can be used to address multipath and multipath-like interference (without loss of generality, hereafter referred to as multipath interference) in two parts: identification and mitigation. A receiver 400 equipped with the invention naturally lends itself to identifying potential multipath interference. Simply, an interference resistant GPS receiver 400 according to the present invention will attempt to acquire and track a second or more fingers (i.e., substantially identical signals that are delayed in time due to, e.g., reflection) for all active GPS processing channels. This technique is similar to those currently being used in the CDMA cell phone domain to acquire and track multipath signals.

[0182] When two channels of the receiver have “good lock” on the same PRN, then multipath interference is occurring. Comparisons of relative signal strengths, the use of PVT solutions using combinations of signals, and/or the use of additional navigation information (e.g., an inertial navigation system (INS)) can be used to identify the true signal and the corresponding true PVT solution. The remaining multipath signals may be eliminated as interference.

[0183] Although the present invention has been fully described in conjunction with the preferred embodiment with reference to the accompanying drawings, it is to be understood that various changes and modifications may be apparent to those skilled in the art. Such changes and modifications are to be understood as included within the scope of the present invention as defined by the appended claims.