GB2213994A - Adaptive antennas - Google Patents

Abstract

An adaptive antenna array comprises a plurality of antenna elements the outputs of which feed a cascaded beamforming network having a succession of stages connected as shown, each stage including a group of signal decorrelation cells (DC), the group in each stage having one less cell than the group of the preceding stage, and includes means for applying weighting to the signals applied as inputs to the cells of at least the first stage. A reference signal being a representation of the desired signal output is applied to the beamforming network as an input additional to the outputs of the antenna elements, and a delayed 40 version of the reference signals subtracted from the output of the last stage. Feedback means (not shown) are provided whereby the difference signal so obtained is used to generate the succeeding reference signal. <IMAGE>

Description

ADAPTIVE ANTENNAS This invention relates to adaptive antennas using cascade beamforming architectures.

Adaptive beamforming provides a powerful means of enhancing the performance of a broad range of communication, navigation and radar systems in hostile electromagnetic environments. In essence, adaptive arrays are antenna systems which can automatically adjust their directional response to null interference or jamming and thus enhance the reception of wanted signals. In many applications, antenna platform dynamics, sophisticated jamming threats and agile waveform structures produce a requirement for adaptive systems having rapid convergence, high cancellation performance and operational flexibility.

In recent years, there has been considerable interest in the application of direct solution or 'lopes loop" techniques to adaptive antenna processing in order to accommodate these increasing demands. In the context of adaptive antenna processing these algorithms have the advantage of requiring only limited input data to accurately describe the external environment and provide an antenna pattern capable of suppressing a wide dynamic range of jamming signals.

The objective of an optimal adaptive antenna system is'to minimise the total noise residue (including -jamming and receiver noise) at the array output whilst maintaining a fixed gain in the direction of the desired signal and hence lead to a maximisation of resultant signal to noise ratio.

According to the present invention there is provided an adaptive antenna array comprising a plurality of antenna elements the outputs of which feed a cascaded beamforming network having a succession of stages, each stage including a group of signal decorrelation cells, the group in each stage having one less cell than the group of the preceding stage and the first stage group having one less cell than the number of antenna elements, each cell of the first stage having as one input the output of a respective antenna element and as a second input the output of the remaining antenna element to produce an output signal and each cell of each subsequent stage having as one input the output of a respective cell of the preceding stage and as a second input the output from the remaining cell of the preceding stage to produce an output signal, the whole arrangement including means for applying weighting to the signals applied as inputs to the cells of at least the first stage, characterised in that means are provided for generating a reference signal being a representation of the desired signal output, each stage including a further decorrelation cell which in the first stage has as one input the reference signal and as a second input the output of the remaining antenna element and the further cell of each succeeding stage having as one input the output of the further cell in the preceding stage and as a second input the output from the remaining cell of the preceding stage, means for subtracting a delayed version of the reference signal from the output of the last stage and feedback means whereby the difference signal so obtained is used to generate the succeeding reference signal.

The invention will now be described with reference to the accompanying drawings, in which: Fig. 1 illustrates an adaptive antenna arrangement to allow the inclusion of reference signals, Figs. 2(i) and (ii) illustrate circuits for reference signal generation, Fig. 3 illustrates a simplified adaptive antenna arrangement using a spread spectrum reference signal, Fig. 4 illustrates a sequential decorrelator arrangement incorporating a reference signal, Fig. 5 illustrates a systolic QR processor incorporating a reference signal, and Fig. 6 illustrates acquisition and data detection performance using a systolic QR processor.

Consider the scheme shown by Figure 1. Signals from the antenna elements 10a-l0n are weighted and then combined in combiner 11 with the reference signal, r(t).

The adaptively controlled weights are adjusted to minimise the mean square residual from the combiner 11 corresponding to the minimum error residue between the reference and the combined element signals. The requisite beamformed response is obtained by subtracting the reference signal from the error residual.The residual from the combiner is: e(t) = wTx(t) + r(t) ... (1) and the mean square of the residual is represented by * T P = E e (t) e(t) = E E(w X(t) + * = E T0 r(t)) (w x(t) + r(t)) - = E [WHX* (t) xT (t) W) + E [WHX* (t) r(t) + E [r (t) X (t) W + E [r*(t) r(t) = WH E Ex (t) xT (t)] W * + / E [X* (t) r(t) + E Er (t) XT(t)] W + E [r*(t) r(t)] ...(2) where E[.] denotes the expectation operator.

The complex gradient of PO is: #Po= 2. [E [X* (t) XT (t)]. W - E [X* (t). r(t))]] ...(3) and the optimal weight solution is obtained when #PO = O and is therefore given by: E CX (t) XT(t)] W = E Ex* (t) r(t) ... (4) This weight solution when applied to the antenna element signals produces a resultant beamformed response which is the closest replica of the injected reference in the least-squares sense.

The beamformer shown in Figure 1 could be controlled adaptively by a closed loop processor which attempts to minimise the error residual by a gradient descent algorithm This technique is known to have a poor convergence characteristic under adverse signal and jamming conditions. Alternatively, the weight vector could be solved by an "open-loop".approach where Maximum Likelihood estimates of the covariance components are used instead of the values contained in the matrices * T * E Ex (t) X (t) and E CX (t) r(t). The direct solution vector is then given by: (XnHXn)W = XHnr ... (5) where Xn and r contain n time samples of X(t) and r(t) respectively.

It is well known that the solution to equations (4) and (5) can be obtained by cascade adaptive beamforming architectures. For example, the general type of problem described by equation (4) is solved in the -steady-state by the Sequential Decorrelator, originally proposed by McQueen, J.G., "Adaptive Cancellation Arrangement", UK Patent Specification No. 1,599,035.

Alternatively, the true least-squares solution of equation (5) can be solved by a data domain algorithm known as QR decomposition. This latter method can be implemented in an efficient, pipelined systolic array.

For the generation of the reference signal consider that the desired signal is bi-phase modulated by a direct sequence outer PRN code and an attempt is made to remove this at the array output by the injection of a locally generated code, as shown by Figure 2(i). If the local code is in perfect alignment with the received code the desired signal at the output from the mixer 20 will be compressed in bandwidth, from that appropriate to the code chipping rate to that of the underlying data modulation. The mixer output may therefore be passed through a narrow bandpass filter 21. The width of this filter is constrained by the bandwidth of the data modulation or the uncertainty in the Doppler shift of the received carrier, whichever is the greater. If the data modulation also takes the form of biphase shift keying the desired signal will have a constant amplitude at this point. It may then be fed to a bandpass limiter 22 and remodulated by the local PRN code to produce a suitable reference signal.

Initially there will be a timing offset between the locally generated and received codes. We must therefore slip the local code relative to the received code until the correlation occurs. When the two codes are misaligned by more than one chip there will be no effective correlation, and the output from the reference generation circuit will be meaningless. However, as soon as the two codes are within one chip of perfect alignment the output signal from the reference signal generation circuit will begin to take on the correct form to assist the convergence of the adaptive array. Provided that the -local and received codes are not slipping past each other at too fast a rate the adaptive algorithm should move appreciably towards convergence before the codes pass out of alignment again.Provided this leads to an adequate signal-to-noise ratio for detection of the correlation we may discontinue the code slip. Thereafter the local and received codes may be kept in alignment with the aid of a code tracking loop.

Provided the jammers do not have a knowledge of the outer PRN code they will not be able to capture the reference signal generation circuit in preference to the desired signal. This is because, even if the jammer is of constant amplitude, the effect of the code injection mixer and subsequent narrow bandpass filter will be to induce amplitude fluctuations in the jamming waveform.

This waveform will therefore have a time-dependent amplitude at the input to the circuit designed to preferentially select the component with constant amplitude.

Figure 3 shows a simplified block diagram of the total system. Here the antenna signals and reference signal are applied to the adaptive combiner 30. A delayed version of this reference is subtracted from the error residual produced at the output of the combiner and the resulting signal corresponds to the "beamformed" response. This is used to (i) generate the next reference signal sample and (ii) aid code generation and tracking.

The delay 31 imposed on the reference signal is necessary due to the throughput or pipeline delay through the adaptive processor. This is usually of the order of 2N data samples where N is the size of the array. The throughput delay also means that a forward prediction of the locally generated PRN code must be used within the reference generation circuit (see figure 2ii) so that, once acquired, the timing of the injected reference signal will be in exact alignment with the received -code. Because we are now in effect using "old" data modulation to estimate the current desired signal, we must ensure that the time constant of the adaptive process is significantly long to prevent adaptation within a data bit period. Figure 2(ii) shows the revised spread spectrum reference generation circuit for use with adaptive beamforming architectures having an appreciable throughput delay.The code generator 23 must now provide a punctual code P and an advance code A.

It can be shown that the Sequential Decorrelator adapts to a steady-state condition which provides an effective weighting transformation identical to the linearly constrained adaptive combiner illustrated in Figure 1.

Figure 4 shows a simplified block diagram of a Sequential Decorrelator incorporating a reference signal. The reference is applied to the right hand input and the process of the decorrelation cells DC produces, at steady-state, a minimised residual power from the beamformer. A delayed version of the reference signal is then subtracted from the network output to obtain the desired beamformed response. The additional delay 40 is matched to the throughput delay imposed by the Sequential Decorrelation process.

A systolic QR processor arrangement incorporating a reference signal is shown in Figure 5.

As with the Sequential Decorrelator, the reference signal is applied to the right-hand input and then a delayed version is subtracted from the error residual sequence.

The delay 50 must be matched to the pipeline delay through the systolic network.

Figure 6 illustrates the code acquisition performance of this scheme obtained by computer simulation. In this example, the desired signal was modelled as- a biphase, PRN sequence with random data modulation applied at a rate some 200 times slower than the chipping rate of the direct sequence outer code. The local code generator was initially set 2 chips out of alignment and then slipped at a rate of 0.001 chips/ sample to allow the tracking loop to lock. A 2 element adaptive antenna has been considered receiving signals from the desired source at a relative strength of -30 dB and a single jammer at O dB. The jammer and desired signal were spaced by 350 in angle.

The top curve shows the misalignment of the locally generated code as the beamformer adapts and the code tracking loop pulls in. The lower curves indicate the transmitted data sequence and the detected data sequence. In this example, the tracking loop bandwidth and sliprate were selected such that code acquisition would not be obtained without the adaptive antenna process.

Claims (4)

CLAIMS:

1. An adaptive antenna array comprising a plurality of antenna elements the outputs of which feed a cascaded beamforming network having a succession of stages, each stage including a group of signal decorrelation cells, the group in each stage having one less cell than the group of the preceding stage and the first stage group having one less cell than the number of antenna elements, each cell of the first stage having as one input the output of a respective antenna element and as a second input the output of the remaining antenna element to produce an output signal and each cell of each subsequent stage having as one input the output of a respective cell of the preceding stage and as a second input the output from the remaining cell of the preceding stage to produce an output signal, the whole arrangement including means for applying weighting to the signals applied as inputs to the cells of at least the first stage, characterised in that means are provided for generating a reference signal being a representation of the desired signal output, each stage including a further decorrelation cell which in the first stage has as one input the reference signal and as a second input the output of the remaining antenna element and the further cell of each succeeding stage having as one input the output of the further cell in the preceding stage and as a second input the output from the remaining cell of the preceding stage, means for subtracting a delayed version of the reference signal from the output of the last stage and feedback means whereby the difference signal so obtained is used to generate the succeeding reference signal.

2. An adaptive antenna array according to claim 1 characterised in that the array includes signal tracking means to which the difference signal is applied, the output of the signal tracking means controlling the timing of the generation of the reference signal.

3. An adaptive antenna array according to claim 1 or 2 wherein the beamforming network is implemented as a systolic QR processor arrangement.

4. An adaptive antenna array substantially as described with reference to the accompanying drawings.