US 20060030287 A1
A receiver with reduced near field noise is described. The receiver has a far range receiving section configured to sense a desired signal that includes near field noise. The receiver further includes a near range receiving section configured to sense a near field noise reference signal. An adaptive noise canceller (ANC) of the receiver is configured to detect the magnitude of an error vector from the far range receiving section. The ANC is configured to adjust the phase and gain of the near field noise reference signal in response to the magnitude of the error. Accordingly, the ANC can generate a corrected near field noise reference signal that is added to the desired signal with an adder. The near field noise is canceled by the addition of the corrected near field noise signal. The ANC uses a least mean square technique to determine the amount of correction needed.
1. A receiver with reduced near field noise, the receiver comprising:
a far range receiving section configured to sense a desired signal having near field noise;
a near range receiving section configured to sense a near field noise reference signal;
an adaptive noise canceller configured to detect the magnitude of an error vector from the far range receiving section by correlating the sensed desired signal from the far range receiving section with a known noise interference reference signal, and to adjust the phase and gain of the near field noise reference signal in response thereto in order to generate a corrected near field noise reference signal; and
an adder configured to add the corrected near field noise reference signal to the desired signal in order to cancel the near field noise.
2. The receiver of
a far range antenna configured to sense the desired signal and the near field noise;
a far range bandpass filter electrically connected to the far range antenna; and
and a far range amplifier electrically connected to the far range bandpass filter;
wherein the far range antenna, the far range bandpass filter, and the far range ampliflier are configured to sense the desired signal and the near filed noise.
1. Field of Invention
The present invention relates generally to noise cancellation systems and more particularly to an adaptive noise cancellation system capable of reducing received noise.
2. Status of the Prior Art
Considerable radio noise is generated by Personal Computers (PC's), as well as other portable computing devices. The noise created by these devices can interfere with the reception of signals by devices such as Wireless Wide Area Network Adapters thereby reducing the sensitivity of the adapter and hence the range to the base station.
The interference can be reduced by suppressing the noise at the source through improved design of the noise emitting electronic device. Alternatively, the noise can be reduced by choosing an antenna for the receiving device which isolates the antenna from the computer using distance (i.e., remote cable connection) or other means. However, these solutions have not been effective because of the reluctance of device manufacturers to increase product cost and a user's reluctance to use a remote cabled antenna.
A common problem with both PCMCIA and OEM wireless modules is that host generated noise can cause desense of the modem on one or more channels of the wireless data service. Desense refers to host generated Electro-Magnetic Interference (EMI) increasing the effective level of the noise floor and reducing the effective sensitivity of the receiver. Measurements have shown that desense in the laptop environment for the PCS band can be as high as 19 dB and for the 850 MHZ band can be as high as 30 dB.
The desense typically arises from digital clock noise generated by the computing device. The clock noise creates harmonics and other spectral components which lie within the bandwidth of the radio channel being used. If these spectral emissions occur within the channel being used for data communication, then problems of interference can occur. The emissions are strong enough to significantly degrade the input sensitivity of the receiver, even though their strength is low enough to meet regulatory emission requirements.
Most common current paths within an electronic device (such as a personal computer) consist of I/O cables, printed circuit board (PCB) signal traces, power supply cables, and power-to-ground loops. Each of these current paths can function as an antenna which radiates electric and magnetic fields. Interaction of these fields with other signals is EMI. The magnitude of the EMI is a function of several characteristics of the transmitted signal—such as frequency, duty cycle, and voltage swing (i.e., amplitude). Determining the role of transmitted signal characteristics is best analyzed in the frequency domain using a Fourier transformation. Any periodic function in the time domain f(t), may be represented by an infinite series of sines and cosines:
If the signal is non-periodic (such as hardware with a microcontroller which references RAM, Flash, I/O devices, control lines, etc. in a time varying fashion), the Fourier Series representation of f(t) (i.e., Equation 1) would contain terms for a wide range of fundamental components such as fundamental frequencies and all of their harmonics.
In a typical PCMCIA or OEM installation, the signal spectrum near the logic boards would appear to be fairly wideband in nature and comprise a large number of individual spectral peaks whose amplitude would vary in time with the function being performed by the digital logic of the board.
The frequency spectrum generated by the high clock speeds and sharp edges of clocks in modern digital devices can extend well into the GigaHertz region. As such, these signals may be within the allocated bandwidth of commercial communication services. As previously mentioned, these signals may be relatively low in amplitude to satisfy the requirements of regulatory emission levels. However, these signals are quite strong when compared to the Received Signal Strength Indication (RSSI) of wireless network transmissions. For example, the RSSI from a base station may be in the order of about −85 dBm, but the level of interference from nearby digitally generated noise may be in the order of −80 dBm. As is evident, a −5 dBm signal to noise ratio results in this example and would degrade the overall wireless network performance.
The present invention addresses the above-described interference problems associated with electronic devices by providing a system and method of adaptive noise cancellation. In this respect, the present invention provides a system which can adaptively cancel electronic device generated noise from nearby radiating electronic devices.
In accordance with the present invention, there is provided a receiver with reduced near field noise having a far range receiving section that is configured to sense a desired signal having near field noise. The receiver further includes a near range receiving section configured to sense a near field noise reference signal. An adaptive noise canceller (ANC) of the receiver is configured to detect the magnitude of an error vector from the far range receiving section and adjust the phase and gain of the near field noise reference signal in response thereto. Accordingly, the ANC is configured to generate a corrected near field noise reference signal that is added to the desired signal with an adder. The near field noise is canceled by the addition of the corrected near field noise signal. The ANC uses a least mean square technique to determine the amount of correction needed.
The far range receiving section includes a far range antenna, a far range bandpass filter and a far range amplifier which are operative to sense the desired far field signal having near field noise. Similarly, the near range receiving section includes a near range antenna, a near range bandpass filter and a near range amplifier which are operative to sense the near field noise reference signal. In order to generate the corrected near field reference signal, the receiver further includes a phase corrector electrically connected to the ANC and operative to correct the phase of the near field noise reference signal in response to the magnitude of the error vector. Furthermore, the receiver includes a gain corrector electrically connected to the ANC and operative to correct the gain of the near field reference signal.
In accordance with the present invention, the receiver may further include a demodulator electrically connected to the ANC and operative to demodulate the signal therefrom. In order to further process the signal from the far field antenna, the receiver may further include an in-phase path and a quadrature phase path. The in-phase path has a mixer operative to mix the signal from the far field bandpass filter with a local oscillator signal that has been phase shifted by ninety degrees. The in-phase path further includes a low pass filter electrically connected between the mixer and a digital to analog converter (DAC). The low pass filter and the DAC are operative to produce a digital representation of the received signal before processing by the ANC. Similarly, the quadrature phase path includes a mixer to mix the signal from the far range bandpass filter with a local oscillator signal. The signal from the mixer in the quadrature phase path is then passed through another low pass filter and another DAC before being inputted into the ANC.
In accordance with the present invention, it is also possible to correct the phase and gain of the near field noise reference signal using a tap delay line (TDL) which receives compensation coefficient signals from the ANC. Specifically, the ANC generates a gain compensation coefficient signal and a phase compensation coefficient signal in response to the magnitude of the error from the far range receiving section. The gain compensation coefficient signal is mixed with the near field noise reference signal to generate a gain compensated near field noise reference signal. The phase compensation coefficient signal is mixed with the near field noise reference signal to generate a phase compensated near field noise reference signal. Next, the gain compensated near field noise reference signal and the phase compensated near field noise reference signal are added together to generate the corrected near field noise reference signal.
In accordance with the present invention, there is provided a method for reducing near field noise in a desired signal. The method commences by sensing the desired signal having near field noise. Next, a near field noise reference signal is sensed. A compensation signal is then generated with an adaptive noise canceller by detecting the magnitude of an error vector from the far range receiving section. The phase and gain of the near field noise reference signal is then adjusted with the compensation signal in order to generate a corrected near field noise reference signal. Finally, the corrected near field noise reference signal is added to desired signal in order to cancel the near field noise.
These, as well as other features of the present invention, will become more apparent upon reference to the drawings wherein:
Referring now to the drawings wherein the showings are for purposes of illustrating a preferred embodiment of the present invention only, and not for purposes of limiting the same,
Connected to the duplexor 14 is a bandpass filter (BPF) 16 with a pass band for the desired frequencies to be received. The output of the bandpass filter 16 is fed into a low noise amplifier (LNA) 18 which amplifies the received signal. The output of the LNA 18 is fed to an adder 20 which adds the signal detected by the signal antenna 12 with a noise cancellation signal, as will be described in greater detail below. The far range signal antenna 12, BPF 16, and LNA 18 form a far range receiving section of the receiver 10.
The receiver 10 further includes a local oscillator (LO) 32 which generates a local oscillator signal from a phase lock loop (PLL) (not shown). The phase of the LO signal from the LO 32 is shifted ninety degrees by phase shifter 34. The phase shifted LO signal is then mixed with the signal from the adder 20 with mixer 36 in an in-phase (I) path of the receiver. The output of the mixer 36 is inputted into a low pass filter (LPF) 40 which passes the desired low frequencies of the mixed signal. The output of the LPF 40 is inputted into an analog to digital converter (ADC) 46 which produces a digital representation of the signal. After the signal has been converted into a digital representation, the signal is inputted into an adaptive noise canceller 50.
The signal from the adder 20 is also processed through a quadrature phase (Q) path of the receiver 10. Specifically, the output of the adder 20 is inputted into mixer 38 which mixes the signal with the LO signal from the LO 32. The output of the mixer. 38 is passed through LPF 42 and then ADC 48. The digital representation of the signal (in the quadrature phase path) from the ADC 48 is then inputted into the ANC 50. The output of the ANC 50 is inputted into a demodulator 52 for signal conversion.
The ANC 50 is able to measure the magnitude of the error vector for the complex baseband signal from the quadrature and in-phase paths. Then the ANC 50 can adjust the gain and phase of a noise canceller loop to minimize the effects of any error. Specifically, the receiver 10 has a near range interferer antenna 22 which detects any near range interference caused by radiating EMI from the electronic host device. The interferer antenna 22 can therefore obtain a near field noise reference signal which is passed through BPF 24 and LNA 26. The interferer antenna 22, BPF 24 and LNA 26 form a far range receiving section of the antenna 10. Next, the gain of the noise reference signal is adjusted according to a compensation signal generated by the ANC 50.
Specifically, the output of the LNA 26 is inputted into gain corrector 28 which adjusts the gain of the noise reference signal in response to a gain compensation signal generated by the ANC 50. After passing through the gain corrector 28, the phase of the noise reference signal is adjusted by a phase corrector 30 in response to a phase compensation signal generated by the ANC 50. The output of the phase corrector 30 is then inputted into the adder 20 such that the corrected noise reference signal is added to the received signal to minimize interference.
If s(t) represents the desired signal from the signal antenna 12 and n1(t) represents the undesired noise also received from an interferer at antenna 12, then the receiver path signal is:
The interference loop path should ideally receive the undesired noise only and will be scaled and phase shifted due to the BPF 24, LNA 26 and the physical separation between signal antenna 12 and interferer antenna 22. The signal at interferer antenna 22 is:
At the adder 20, the desired signal is added with the scaled and phase shifted noise reference signal to result in:
The goal of the ANC 50 is to minimize the term n1(t)+G*a*n1(t−τ+Φ). By minimizing the error vector magnitude for the desired signal s(t) at baseband, the ANC 50 desires a situation where:
The advantage of this scheme is that once G and Φ are optimized in terms of the error vector magnitude, they do not need to adapt to any changes in the time domain characteristics of the interference. Because only two parameters are adapted (i.e. G and Φ), it is easy for the ANC 50 to calculate the result using a Least Mean Squares (LMS) equalizer. Alternatively, the adaptation could easily be dealt with by Falconer's algorithm. However, the convergence of the equalizer must also ensure equalizer self-noise doesn't degrade the performance of the demodulator 52. In order for the noise cancellation scheme to work, the noise n2(t) must be highly correlated with n1(t), but it does not need to be known in advance.
In order to ensure optimal cancellation, the two paths (i.e, the received signal path and the noise cancellation path) should be identical as possible. Accordingly, identical BPF's 24, 16 and LNA's 18, 26 should be used. This will ensure that any effects of differential group delay between the desired channel and the interference channel will be minimized.
The noise reference signal from the near range interferer antenna 22 can be derived in various ways. A first method is to create a loop identical to the signal antenna 12 in terms of connectorization and cabeling, except that the loop would be terminated with a short circuit to ground. This method provides a conducted interference reference. Alternatively, H-field antennas could be used to sample the interference. Because the desired signal antenna 12 is an E-field antenna, an H-field antenna would offer rejection of the desired signal which is beneficial for the cancellation process.
In the example shown in
Accordingly, the noise reference signal has an I path wherein the noise reference signal from the LNA 26 is mixed by a mixer 136 with a LO signal from LO 132 that has been phase shifted ninety degrees by LO shifter 134. After being mixed with the phase shifted LO signal, then the noise reference signal is passed through LPF 140 and DAC 146 and into ANC 50. Similarly, the noise reference signal from LNA 26 is mixed with the LO signal from LO 132 by mixer 138 in the Q path. The mixed noise reference signal is then passed through LPF 142 and DAC 148 before being inputted into ANC 50. The ANC 50 is able to separately process the noise reference signal in order to cancel the noise in the desired signal. Accordingly, the ANC 50 can add the noise cancel signal to the received signal in order to cancel noise.
However, by including an RF switch 76 that switches between the near range interferer antenna 22 and a far range interferer antenna 23 it is possible to reduce both far field and near field noise. Accordingly, the circuit 74 includes RF switch 76 which is operable to switch between the near range interferer antenna 22 which senses the near field reference signal and the far range interferer antenna 23 which senses the far field signal plus interference. Furthermore, the circuit 74 includes an ANC and adaptive steering module 51 which controls the switch 76 in order to choose which antenna 22 or 23 is to receive the noise reference signal. The ANC and adaptive antenna steering module 51 is similar to the ANC 50 but further includes the ability to choose and steer the antenna beam as desired. Accordingly, the ANC and adaptive antenna steering module 51 could implement either a phase array beam steering algorithm to minimize far field interference, or an adaptive noise canceling technique to minimize the near field interference, as previously described.
As previously mentioned above, the ANC 50 uses a cancellation technique in order to generate the compensation signal that cancels noise in the desired signal. A general form of the ANC 50 is shown in
The two most common optimization techniques are:
As previously mentioned, the ANC 50 processes the noise reference in such a way as to minimize the cost function. In order for this cancellation to occur, there must be some degree of correlation between the noise reference signal and the noise which is degrading the signal. The most common form to realize the adaptive cancellation technique is the direct form even though other forms are available (i.e., cascade form, parallel form, and lattice form). A system block diagram for a direct form conventional FIR filter is shown in
The output y(k) of the filter shown in
In equation 14, g(k) is the key to the convergence because it represents the estimated gradient or the projection of the square of the current error signal 2e(k) on the filter tap weight. When the algorithm converges, g(k) is expected to be a very small number with zero mean value.
The step size u in Equation 13 must be in the range of 0<u, I/Lmax where Lmax is the largest Eigenvalue of the covariance matrix R. R is given by the equation:
In practice, u should be much smaller than 1/Lmax when Lmin is much smaller than Lmax. The number of steps it takes for the algorithm to converge is proportional to Lmax/Lmin.
An example of how the adaptive noise cancellation technique removes noise will now follow. A 5-tap FIR LMS equalizer was used to remove a single sine wave interferer from a signal source. A basic binary signaling scheme (i.e., Pulse Amplitude Modulation) was used as the desired signal. A single sine wave interferer with variable amplitude and variable noise was added to the baseband signal. The step coefficient was adjustable to allow the optimum value to be determined empirically. In the example, the uncorrupted baseband signal had values of ±1 and the sine wave ranged in magnitude from −10 to +10. Accordingly, a signal to noise ratio (SNR) of approximately −20 dB was generated. Random noise with a RMS value of about 0.025 was also added to the signal. The uncorrupted sine wave was used as the noise interference reference signal.
When a simple LMS equalizer is applied to the baseband signal corrupted by a single CW interferer (
An example of multiple CW interferers will now be discussed. For this example, the second interferer will have a magnitude of half that of the first interferer and a frequency of 3.1456 times the first interferer. The combined interference magnitude is limited to the range of −10 to +10. A plot of the corrupted baseband signal with multiple CW interferers prior to noise cancellation is shown in
In the case of multiple interferers, considerable improvement is achieved after the application of the techniques used by the ANC 50. A plot of the mean square error is shown in
Additionally, a time varying single CW interferer is shown in
In addition to the foregoing, it is also possible to use the adaptive noise cancellation technique of the present invention even though a noise reference signal is not available. In this instance it is possible to use data embedded in the received signal as a reference to aid in the noise cancellation process. Referring to
A difficulty arises by the effect of large interference on the correlation coefficient. Referring to
Specifically, the 35 bit synch word 80 consists of 7 groups of 5 bits which can be decoded differentially as 7 groups of 4 bits. The corrupted signal is differentially decoded and then cross correlated with a 28 bit reference function. The forward channel synch word 80 is not changed, but is treated as a differentially encoded codeword, and then differentially decoded and correlated with the differentially decoded version of the 35 bit synch word 80. The result is shown in
As is evident, by utilizing the forward channel synch word it is possible to adapt the noise canceller rather than doing purely blind cancellation using estimates of the symbol decisions. Referring to
It will be recognized that although the above-described illustration used the CDPD forward channel as an example, similar a priori knowledge of a forward channel signaling scheme for other wireless protocols could be used in a similar fashion. Furthermore, even though the LMS technique is discussed for the previous examples, it will be recognized that other techniques exist. Accordingly, Recursive Least Squares and Kalman techniques that achieve much lower noise floors than the LMS technique can be used. However, these other techniques require considerably more computational power than the basic LMS algorithm.
Additional modifications and improvements of the present invention may be apparent to those of ordinary skill in the art. Thus, the particular combination of parts described and illustrated herein is intended to represent only certain embodiments of the present invention, and are not intended to serve as limitations of alternative devices within the spirit and scope of the invention.