Space-Multiplexed Time-Integrating Acousto-Optic Correlators

1. Field of the Invention

The present invention relates to the field of acousto-optic devices and acousto-optic signal processing.

2. Prior Art

Acousto-Optic (acousto-optic) devices are well-known and widely used light modulators, being generally described in the literature, including Proc. IEEE, Special Issue on Acousto-Optics, Vol. 69, January 1981, and Acousto-Optic Signal Processing: Theory and Implementation, Ed. N. J. Berg and J. N. Lee, Marcel Dekker, Inc., New York, 1983.

In accordance with this technology an input electrical signal s(t) to such a device is converted to a sound field in the acousto-optic cell crystal by an input transducer. This wave then travels the length of the crystal, with an absorber at the far end of the device causing the wave to terminate at the end of the device with no reflections. The input electrical signal is presented on a carrier as s.sub.1 (t)=s(t) cos .omega..sub.c t or s.sub.2 (t)=[B +s(t)] cos .omega..sub.c t, where s(t) is a zero-mean signal and B is a bias. When illuminated with light, the cell diffracts the input light at angles proportional to n.omega..sub.c. These waves are referred to as diffracted orders, and the wave .varies..+-..omega..sub.c as the first order wave.

As the sound field travels the length of the cell, the sound field s(x,t) in the cell varies in space x and time t. Depending on the acousto-optic cell and the input signal s.sub.1 (t) or s.sub.2 (t), the amplitude or intensity of the first-order wave can be made proportional to s(t) or B+s(t) respectively. For amplitude modulation, the input electrical signal is s(t) cos .omega..sub.c t and the amplitude of the first-order wave is

i.e. the amplitude is proportional to s(t-x/v)

where K is a constant, A.sub.in is the amplitude of the input light wave and .omega..sub.L is its frequency, and v is the velocity of sound in the acousto-optic material. For intensity modulation, the input electrical signal is [B+s(t)] cos .omega..sub.c t and the intensity of first-order wave is

where K is a constant and I.sub.in =.vertline.A.sub.in .vertline..sup.2. Thus, except for a constant bias, the intensity is proportional to s(t-x/v),

By a single change of variables, (2) and (4) can be written as s(x-vt). The representations in (2) and (4) are more appropriate for a time-integrating acousto-optic processor as shall subsequently be seen.

The classic time-integrating acousto-optic correlator of FIG. 1 is well-known and described in detail elsewhere, including the two references previously referred to and in R. A. Sprague and C. L. Koliopoulos, "Time Integrating Acousto-Optic Correlator", Applied Optics, Vol. 15, pp. 89-92, January 1976; and P. Kellman, "Time Integrating Optical Processors", in Optical Processing Systems, W. Rhodes, ed. (Proc. SPIE, Vol. 185, 1979), pp. 130, 1979. Ignoring Bragg or Raman-Nath mode, amplitude or intensity modulation, any bias and .omega..sub.c carrier, and single-sideband filtering (described in the foregoing references), the operation of the system can easily be described. The system of FIG. 1 consists of a point modulator fed with a signal s.sub.b (t). Its output is expanded (by lens L.sub.1) to uniformly illuminate an acousto-optic cell at P.sub.2. The light distribution incident on P.sub.2 is thus s.sub.b (t), varying in time and being uniform in space. With s.sub.a (t) fed to the acousto-optic cell, its transmittance is s.sub.a (t-.tau.), where .tau.=x/v as in (2) or (4). The light leaving P.sub.2 is now s.sub.b (t)s.sub.a (t-.tau.). Lenses L.sub.2 image P.sub.2 onto P.sub.3 (and SSB filters the result). Since any bias and the .omega..sub.c carrier have been ignored, the pattern leaving P.sub.2 and the pattern incident on P.sub.3 are the same. The detector at P.sub.3 time integrates the incident pattern and the P.sub.3 output obtained is

i.e. the correlation (symbol ) of s.sub.a and s.sub.b is displayed as a function of space (.tau..varies.x) at P.sub.3.

The time integrating correlator is advantageous when T.sub.S >T.sub.A and TBWP.sub.S 22 TBWP.sub.A, where T.sub.s is the signal duration, T.sub.A is the acousto-optic cell aperture time, TBWB.sub.S is the signal time-bandwidth product and TBWP.sub.A is the acousto-optic cell time-bandwidth product. The processor of FIG. 1 can thus provide the correlation output for a very long signal, with the integration time T.sub.I of the detector determining the T.sub.S =T.sub.I value used. If detector dynamic range is exceeded, the contents of the detector are dumped and stored (after some T.sub.I '<T.sub.S) and a new integration is begun. By noncoherently adding the R(.tau.) outputs for separate .tau..sub.I ', the full T.sub.I =T.sub.S integration is achieved (at a loss of about 3 dB in processing gain due to the noncoherent summation). The time integrating correlator can however only search a limited time delay between signals T.sub.D (-T.sub.A /2<T.sub.D <T.sub.A/2) set by T.sub.A of the acousto-optic cell, i.e., T.sub.D <T.sub.A.

The purpose of the present invention is to provide a system which can achieve multiple signal correlations and an infinite T.sub.D range delay search.

Space multiplexed time integrating acousto-optic correlators are disclosed. These time integrating processors can achieve multiple signal correlations with an infinite range delay search. In some embodiments, two stage synchronization may be achieved and N channel demodulation may be achieved with automatic sync realignment. Various embodiments are disclosed.

FIG. 1 is a block dagram of a typical prior art time integrating acousto-optic correlator.

FIG. 2 is a block diagram illustrating an embodiment of the present invention which can achieve multiple signal correlations and an infinite range delay search.

FIG. 3 is a block diagram of an alternate embodiment of the system of FIG. 2 to provide a fine sync capability.

FIG. 4 is a block diagram of an embodiment of the invention for a combined sync and demodulation application.

The present invention time integrating processor can achieve multiple-signal correlations and an infinite (T.sub.D =.infin.) range delay search. This new architecture is shown in FIG. 2. In this case, N inputs at P.sub.1 and N detectors at P.sub.3 are used. Lens system L.sub.1 collimates each P.sub.1 source horizontally (to uniformly illuminate the acousto-optic cell at P.sub.2) and focuses all P.sub.1 point modulators vertically to illuminate the acoustic column in the acoutso-optic cell at P.sub.2. L.sub.1 is thus two cylindrical lenses. Denoting the N inputs at P.sub.1 by s.sub.bn, the N waves leaving P.sub.2 are s.sub.bn (t)s.sub.a (t-.tau.). Lens system L.sub.2 collimates vertically (i.e., L.sub.1 and L.sub.2 image vertically) and integrates (Fourier transforms) horizontally. Thus, L.sub.2 consists of two cylindrical lenses. The horizontal lens compresses the light leaving P.sub.2 into the desired horizontal size at P.sub.3. The vertical L.sub.2 lens together with the vertical L.sub.1 lens image P.sub.1 onto P.sub.3 with the desired vertical scale to match the input point modulators and the output detectors.

The system of FIG. 2 thus yields N spatially-separated correlation outputs, with the horizontal size of each controllable. Considered herein are two uses of this system with different s.sub.bn input signals and slightly different detector arrangements. Then considered is a general unified system for both uses.

Consider the case when the N inputs to P.sub.1 are delayed versions s.sub.b1 (t)=s.sub.b(t), s.sub.b2 (t)=s.sub.b (t-T.sub.A), s.sub.b3 (t)=s.sub.b (t-2T.sub.A), etc. of the same reference signals s.sub.b (t) as in FIG. 4. The N correlation outputs are the correlations of s.sub.a with different delayed versions (T.sub.A, 2T.sub.A, 3T.sub.A, etc.) of s.sub.b. If we select

and integrate each output for T.sub.I =T.sub.S, then the full set of N correlations covers a delay

Each T.sub.I, these N correlation outputs are analyzed and the full processing gain over T.sub.I is achieved for a delay T.sub.D =NT.sub.A. The N references at P.sub.1 are cyclically repeated each T.sub.I =T.sub.S. Thus, in the next T.sub.I =T.sub.S, the next T.sub.D =NT.sub.A delays are checked. Thus, this system achieves an infinite T.sub.D delay for signals with T.sub.I =T.sub.S =NT.sub.A.

At each of the N detector locations in P.sub.3, a linear horizontal detector array with M=TBWP.sub.A detector elements can be placed. By reading out these N linear detector arrays, fine sync within T.sub.A /TBWP.sub.A can be achieved (after a delay equal to the readout time of each one dimensional detector array).

A preferable arrangement in many cases would employ only one detector in each of the N correlation locations in P.sub.3. Each detector would be large enough to fully cover each of the correlation planes (the horizontal component of lens L.sub.2 can be chosen to achieve the required horizontal imaging compatible with the detector geometry). In this case, only N detectors need be read out. This can be achieved in parallel each T.sub.I =T.sub.S. The detector with a peak above threshold denotes the delay between s.sub.a (t) and s.sub.b (t) and hence the sync information. However, this is coarse sync within T.sub.D =T.sub.A only. By summing each full correlaion pattern onto one detector, some loss in P.sub.D (probability of detection) and P.sub.E (probability of error) results. A statistical analysis for the specific sync code used can determine the optimum N, T.sub.A and M (number of detectors per correlation plane) values to use for each case. Of course N and T.sub.A must still be chosen to satisfy (6) and/or (7). With M detectors in each of the N output correlation locations, coarse sync within T.sub.A /M results. A typical value might be M=3 as discussed later. After coarse sync within T.sub.A or T.sub.A /M (as described above), s.sub.a (t) can be delayed by T.sub.I =T.sub.S and s.sub.b (t) synchronized within T.sub.A and fed to one (e.g. the central) of the modulators at P.sub.1. On one output channel (e.g. the center of P.sub.3), a fully populated detector with M=TBWP.sub.A elements can be placed. The location of the correlation peak now provides fine sync within T.sub.A /TBWP.sub.A. In practice, a third lens L.sub.3 can be used and the fully populated fine detector with M elements can be placed in a new P.sub.4 plane behind this L.sub.3, as shown in FIG. 3.

To summarize the steps in the two step coarse/fine synchronization, the N delayed S.sub.b (t) signals are applied to P.sub.1, with the N (or NM) detector outputs providing coarse sync within T.sub.A (or T.sub.A /M). Thereafter S.sub.a (t) can be delayed by T.sub.I =T.sub.S and the one S(t) P.sub.1 input spaced within T.sub.A (or T.sub.A /M) applied to the light source for the fully populated channel. Now the fully populated outputs on this one channel provide fine sync within T.sub.A /TBWP.sub.A.

For a demodulation application, the reference signals s.sub.bn (t) are in sync with s.sub.a (t). To demodulate N possible codes, the N references s.sub.bn in FIG. 2 are simply the N codes. The correlation output with a peak above threshold defines which of the N codes is present each T.sub.S =T.sub.I. For demodulation T.sub.S =T.sub.I is generally much less than for synchronization. Electronic control of the detector integration time is easily achieved. With fully populated detector arrays at P.sub.3, only the central element of each need be interrogated (since the reference and received signals are in sync). To allow for drift in sync, M=3 detectors in each of the N correlation outputs can be used. If detection occurs on other than the central detector, the system sync is readjusted.

M=3 detectors covering each of the N full correlation planes appears to provide acceptable detection performance. This simplifies the output detector system, the output plane electronic support and its analysis required. It also allows for a combined system as discussed below.

The synchronization and demodulation systems described above can be combined in various ways. The N input signals to P.sub.1 are electronically controlled to be N delayed versions of the synchronization waveform (FIG. 4) or N codes (FIGS. 2 and 3), depending upon the operating mode (synchronization or demodulation). N fully populated detector arrays with TBWP.sub.A detectors in each are possible. Preferable arrangements utilize M detectors per correlation output (M=3 typically). With N output detector arrays with M elements each and one output detector array with TBWP.sub.A /M detectors, two-stage synchronization (coarse/fine) is possible, N channel demodulation is achieved (FIG. 2), and sync realignment (with M=3 detectors per channel) is achieved. Obviously, while the present invention has been disclosed and described with respect to certain preferred embodiments, it will be understood that various changes in the form and detail may be made therein without departing from the spirit and scope thereof.