Communications: directive radio wave systems and devices (e.g. – Directive – Beacon or receiver
Reexamination Certificate
2002-12-17
2004-07-20
Buczinski, Stephen C. (Department: 3662)
Communications: directive radio wave systems and devices (e.g.,
Directive
Beacon or receiver
C342S195000, C342S421000, C342S445000
Reexamination Certificate
active
06765532
ABSTRACT:
FIELD OF INVENTION
This invention relates to the detection of signals in interference and more particularly to an improved spatial acquisition system utilizing signal subspace tracking.
BACKGROUND OF THE INVENTION
The purpose of a spatial acquisition process is to detect signals in a cluttered RF environment. The ability to ascertain when a new signal birth occurs is extremely important in modern warfare to quickly identify the signal type and the location of its transmitter. Finding signals and identifying them is particularly troublesome when listening in on a wide range of frequencies, so called wideband signal acquisition.
In addition to the problem of discerning a signal from background environment noise, one is faced with the problem of detecting signals occupying the same frequency band, so called co-channel interferers. Single antenna (non-spatial) detection systems attempt to detect new signal energy through changes in amplitude in narrowband frequency channels (i.e. a channelizing receiver). Attempts to detect co-channel signals require that the new signal power exceeds the existing signal level by a significant amount (Signal on Signal detection) in order to trigger a new signal detection. Obviously, this requirement for increased signal power only works in a limited number of situations, and often the signal of interest will be received in a channel at lower power than an existing interferer, making it difficult if not impossible to detect.
It will be shown that by using an array of antennas and processing in a spatial manner, one is able to discern the existence of multiple signals using detection via rank mechanisms and angle of arrival. This adds an additional dimension other than amplitude and allows one to separate the signals in a manner that one cannot do by using a single antenna. As an example, many military radios happen to also occupy the same frequency band as TV stations. If one uses an aircraft for surveillance and one uses a single channel acquisition system, ie one with a single antenna, one is going to detect or see the TV station because it is the stronger of the two signals. In a battlefield scenario the military signal might be hidden amongst the frequency content of the TV station. On the ground military personnel can talk back and forth because they do not receive the signals from the TV station, but in the air one might detect signals from both of these sources. It is therefore important to be able to spatially separate the military signals from the TV signals so as to identify where the signals came from.
Prior Approaches
A prior approach devised by Brian G. Agee of AGI Engineering Consulting, provided beamforming on new signal births or signals which have burst characteristics, off/on. The beamformer ‘weights’ are calculated by looking at the so-called ‘change matrix’ or the new covariance matrix versus the old covariance matrix and generates the beamformer weights which maximize the Signal to Interference ratio for the ‘new’ signal. Therefore, a wide bandwidth detection system based on this approach, and assuming frequency channelization, would require application of this technique on all frequency channels all the time and looking at the output of the beam former to determine whether new energy had arrived at the array. If the amount of output power from the beamformer exceeded a threshold, Agee's process would declare that occurrence a new detection.
A weakness of the Agee process is that the focus is strictly on new energy detection. It considered the old signal environment to be background interference, and simply looked for changes to the background. The Agee process only allowed for a change of one signal or a rank change of one to this background. This prevents for instance deciding that there are two or more new incoming signals. Unfortunately with that approach there was never an understanding of what the background environment was. For instance, signals that were always up would be ignored because they were considered part of the background environment. In other words the Agee process was optimal for new energy alarms or for bursty signals, ie. signals that go off to on and vice versa. However, the Agee process had little benefit for tracking existing signals and was limited to leading edge clustering, only. As a result, to build a more robust reconnaissance system, one needs to not only be interested in signals that are bursty or new but also to be cognizant of the signals that are already in the environment. In order to know the number of signals present, both old and new, one required a separate processing path to look for the signals that were stable, e.g. the ones that were not bursty in nature.
In short, assuming that multiple signals arrived at an antenna array, in the past there was no particularly good way of ascertaining how many steady state signals were there as opposed to bursty signals. What this means is that if for instance there were four signals that were constantly on and two more signals arrived, the Agee process would associate the four signals with the background environment, i.e., noise, and associate the two new signals as a combined signal in the signal subspace.
Determining the Number of Signals Impinging on the Array
Therefore, one of the key requirements for the detection of a signal in a dense RF environment is to determnine the total number of signals that are impinging on the array. Existing block processing techniques focus on eigen-space decomposition of the data covariance matrix, then, applying rank estimation techniques such as Minimum Data Length, MDL or Akaike Information Criteria, AIC to separate the signal and noise subspace components. The rank of the signal subspace is equivalent to the number of spatially separated signals impinging on the array. Standard algorithms for performing eigen decomposition include the symmetric QR algorithm and Jacobi rotations. A limiting assumption of these signal subspace decomposition techniques is that the total number of signals impinging on the array is less than the number of array elements or degrees of freedom, DOF. These block processing approaches to signal subspace decomposition have been developed for tasked DF systems, that is to capture a buffer full of data and then post-process the buffered data to determine how many signals are in the buffer. Thereafter one can employ a superresolution DF algorithm, such as the MUSIC algorithm, to determine angle of arrival information.
Articles relating to Superresolution DF and subspace tracking are as follows: Multiple Emitter Location and Signal Parameter Estimator by Ralph O. Schmidt, IEEE Proceedings of the RADC Spectrum Estimation Workshop, October 1979; Adaptive Rank Estimator for Spherical Subspace Trackers by Aleksander Kaveie and Bing Yang, IEEE Transactions on Signal Processing, Vol. 44, No. 6, June 1996; An Extension of the PASTd Algorithm to Both Rank and Subspace Tracking, Bin Yang, IEEE Signal Processing Letters, Vol. 2, No. 9, September 1995; and, Two Algorithms For Fast Approximate Subspace Tracking by Edward C. Read, Donald W. Tufts and James W. Cooley, IEEE Transactions on Signal Processing, Vol. 47, No. 7, July 1999.
Extension to Wide Band Signal Acquisition
For wide bandwidth signal detection across a multiple antenna array, spatial acquisition requires N antennas [e.g., N=8 antennas]. Theoretically, the largest rank or number of signals that can be found with 8 antennas is 7 signals, the other one being the noise signal. With 8 antennas there are eight degrees of freedom. In a wideband system all degrees of freedom can easily be occupied. This is because in a wideband application for instance having a bandwidth of 80 MHz the likelihood that there will be more than 7 signals is virtually guaranteed. What one needs to do is to break up the large band into smaller channelized frequency bands such that the likely number of signals in each of those smaller frequency bands is much less than the full eight, preferably four or less for most practical arrays. The narrower
Paul Norman D.
Vaccaro Thomas R.
Bae Systems Information and Electronic Systems Integration Inc.
Buczinski Stephen C.
Long Daniel J.
LandOfFree
Wideband signal detection and tracking system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Wideband signal detection and tracking system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Wideband signal detection and tracking system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3197121