The present invention relates to a digital receiver adapted for receiving MSK modulated signal. The invention, more particularly, relates to a digital receiver for use in a magnetic induction radio used in a hearing assistive device, and comprising a phase compensation unit adapted for compensating for a phase offset for the samples of a down-mixed signal. Also, the invention relates to a method for compensating for a phase offset for the samples of a down-mixed signal.
In a wireless system, the receiver has no access to the transmitters carrier signal. Hence, the transmitter and receiver will modulate and demodulate with separate frequencies that intentionally are close, but not identical. The frequency difference translates to a phase offset between the transmitted IF and received IF signal. This phase offset varies in time. Optimal coherent detection requires the receiver to detect and compensate for this phase offset.
The purpose of the invention is to provide a receiver detecting and compensating for a phase offset due to difference in the carrier frequency of the received signal and the demodulator.
This purpose is achieved by a method enabling a digital receiver to detect the phase offset for an MSK modulated signal. The method is economic and robust and does not require correctly detected data for proper detection. The invention is defined in claims 1 and 8. Preferred embodiments are defined in the dependents claims.
According to a first aspect of the invention there is provided a digital receiver adapted for receiving an MSK modulated signal and comprising a digital front-end unit adapted for providing samples having a phase value of a down-mixed signal, a phase compensation unit adapted for compensating the phase value by delivering a phase offset compensated sample having a phase value, and a coherent demodulator adapted for recovering information content from the phase offset compensated sample. The phase compensation unit is adapted for analyzing a phase value of the phase offset compensated sample, calculating a phase offset value based on the phase value of the phase offset compensated sample, and applying the phase offset value when delivering a subsequent phase offset compensated sample.
The invention will be described in further detail with reference to preferred aspects and the accompanying drawing, in which:
In the digital receiver 13, complex IF samples are handled at a constant sampling rate in the form of a real and an imaginary sample. The complex IF samples makes it easy to perform the calculation of (θ modulus π/2>π/4) extremely economical. The complex IF input sample is denoted as RE+j IM the calculation whether 0 modulus π/2>π/4 can be done by simple logic decisions based on whether (|IM|>|RE|), (RE>0), and (IM>0). This is illustrated in
Minimum Shift Keying (MSK) is a signaling format where digital information is modulated by changing the phase of a radio signal relative to a fixed carrier frequency. The phase changes at a constant rate (equivalent to a constant frequency deviation relative to carrier) for the duration of the time it takes to transmit one symbol. The phase change is ±90 degrees depending of the digital information to be transmitted. The digital information can for example be coded as +90 degrees for a logic “1” and −90 degrees for a logic “0”. This is explained in more details below.
A continuous phase FSK (CPFSK) signal have the form:
Φ=A cos(ωct+γ(t)),
with the phase γ(t) being a continuous function of time. The signal is based on two frequencies ω1 and ω2, where ωc=(ω1+ω2)/2 and Δω=(ω1−ω2)/2.
In the interval 0<t≤T the phase is a linear function of time:
γ(t)=±Δωt+γ(0),which gives:
Φ=A cos(ωct±Δωt+γ(0)),
For minimum-shift keying (MSK), we have Δω=π/2T. This is the minimum frequency spacing between ω1 and ω2 that allows the two FSK signals to be orthogonal to each other. T is the symbol width. This criterion means that the frequency separation between f1 and f2 must be such that there is a half cycle difference in one-bit interval.
By using this condition, we get:
γ(t)=±tπ/2T+γ(0), for 0<t≤T.
By choosing γ(0)=0, the possible values of γ(t) for t>0 can be shown as a phase trellis diagram. The phase at multiples of T can therefore only assume a set of discrete values. More specifically, over each symbol duration, T, the phase of the MSK waveform can only be advanced or retarded by exactly 90 degrees—it ramps up by 90 degrees when a “1” is transmitted, and down by 90 degrees when a “0” is transmitted.
In a real and non-ideal setup, the phase trellis shown in
When analyzing the phase distribution for samples of the bandwidth-limited signal, it is observed that the distribution of samples will have higher density for phases around π/4+nπ/2, where n is an integer. Furthermore, when inspecting the distribution of samples for the phase in the range 0−π/2, the phase distribution will have center of gravity at π/4 and will be almost symmetrical around the center of gravity. A histogram for a perfectly synchronized signal will be symmetric around π/4 radians. In presence of an undesired phase offset, the histogram will be shifted/rotated right or left depending on the sign and magnitude of offset. The “phase modulus π/2” value of the IF samples is illustrated in
In a digital receiver it is economical affordable to calculate the “phase modulus π/2” value of the IF samples, and to determine whether the result is below or above π/4. This calculus may result in a 2-bin histogram as shown in
The phase offset is generated by an error caused by a mismatch between the carrier frequencies used at the transmitter side and the receiver side, even though the two carrier frequencies ideally should be identical. When the mismatch between the two carrier frequencies remains substantially stable over time, it has been realized that the direction of the phase offset is important. An ideal situation will be when the phase offset is zero. By adding a compensation value to the IF samples, it is possible to compensate for the phase offset caused by a mismatch between carrier frequencies used at the transmitter side and the receiver side.
Even though the presence of noise affects the phase offset for the IF samples, and the probability difference is less significant compared to a noise free IF sample signal, the probability difference can still be used to compensate for phase offset. Data for the distribution of the “phase modulus π/2” value, is according to the invention used in an iterative process where the incoming IF sample signal is adjusted for a phase offset in order to obtain a distribution of the “phase modulus π/2” value being symmetric around π/4.
After the phase compensation, the sample is output in step 22 to the coherent demodulator 12 for symbol detection. Depending on the phase, θsync, of the synchronized signal, a small adjustment is performed to the phase offset, ° offset, by either adding or subtracting an adjustment value θadjust. In step 23, a phase value, θmod, is calculated as the “modulus π/2” value of the phase, θsync, of the synchronized signal.
In step 24, it is evaluated whether phase value, θmod, is above or below π/4. In step 25, after the phase value, θmod, is deemed to be above π/4 in step 24, the phase offset value, θoffset, for the following sample is calculated based on the phase offset value, θoffset, for the current sample as being:
θoffset−θoffset+θadjust.
In case the phase value, θmod, is deemed to be below π/4 in step 24, the phase offset value, θoffset, for the following sample is calculated based on the phase offset value, θoffset, for the current sample in step 26 as being:
θoffset=θoffset−θadjust.
The phase offset value, ° offset, calculated in step 25 or 26 is then used for phase adjusting or synchronizing the next complex sample received in step 20.
The adjustment value, θadjust, is a small predetermined value. In one embodiment the adjustment value, θadjust, is chosen to be below one percent of π/4.
The numeric value of phase offset value, θoffset, will over time reach the point where the probability of a positive adjustment equals the probability of a negative adjustment in step 25 or 26. This equilibrium represents the ideal state, where θsync is correctly adjusted. The phase offset value, θoffset, now corresponds to the observed phase offset of the received complex sample, and in the state with the probability of a positive adjustment equals the probability of a negative adjustment, the algorithm shown in
In one embodiment, the phase compensation unit 11 applies an algorithm using a first order integrator loop for the phase compensation. The integration will completely compensate for a static offset, but a constant phase error will remain if the phase offset is a linear function of time, which is the case when transmitter and receiver frequency are different. To completely compensate for linearly varying phase a second order integrator system is required.
The method discussed above uses phase modulus π/2 for phase-offset compensation, and the method compensate phase offset to nearest integer multiple of π/2. This works fine for a coding method e.g. where “1” corresponds to m/2 phase advance, and “0” corresponds to −π/2.
According to one embodiment of the invention, the method of carrier resynchronization of an MSK signal in a digital receiver and comprising steps of providing samples having a phase value (θmeasure) of a down-mixed signal; compensating the phase value (θmeasure) of the down-mixed signal by delivering a phase offset compensated sample having a phase value (θsync); and recovering information content from the phase offset compensated sample. The method further comprises steps of analyzing a phase value (θsync) of the phase offset compensated sample, calculating a phase offset value (θoffset) based on the phase value (θsync) of the phase offset compensated sample, and applying the phase offset value (θoffset) when delivering a subsequent phase offset compensated sample.
In one embodiment, the applying of the phase offset value (θoffset) for phase offset compensation of a sample, uses a phase offset value (θoffset) calculated by means of the immediately preceding sample. In another embodiment, on value of the phase offset value (θoffset) is used for compensating a plurality of samples, e.g. 16 or 64.
The method according to the invention has several advantages compared to prior art methods. Firstly, it is not required at any point to calculate the phase for incoming samples. This would be the case if for examples using preambles with synchronization information. Phase calculations are normally expensive and requires several CORDIC iterations. CORDIC mathematical techniques are optimized for low-complexity finite-state CPUs. Secondly, it is not necessary to distinguish between preambles, messages or pauses in the signal to correctly synchronize the incoming signal.
The method works even with very poor signal to noise ratios. Even though when synchronization is lost during transmission due to poor signal quality, synchronization will automatically be regained when signal quality improves. Correct clock recovery is not required. In fact, clock recovery and synchronization may work independently of each other.
The method does not require data assistance. Hence, vicious circles where incorrectly detected data degrades synchronization and vice versa does not exist with this method. The required calculation of (θ modulus π/2>π/4) is very inexpensive.
According to one embodiment of the invention, the digital receiver in embedded in a radio for use in an inductive radio link between e.g. two binaural hearing aids for increasing the available bandwidth of the receiver.
This application is a National Stage of International Application No. PCT/EP2020/079548 filed Oct. 21, 2020, claiming priority based on U.S. Patent Application No. 62/932,721 filed Nov. 8, 2019.
Filing Document | Filing Date | Country | Kind |
---|---|---|---|
PCT/EP2020/079548 | 10/21/2020 | WO |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO2021/089322 | 5/14/2021 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
3993868 | Balcewicz | Nov 1976 | A |
4583048 | Gumacos | Apr 1986 | A |
4995052 | Thorvaldsen | Feb 1991 | A |
5164965 | Karaali | Nov 1992 | A |
5274672 | Weiss | Dec 1993 | A |
20060291550 | Wang | Dec 2006 | A1 |
20150222419 | Bachmann | Aug 2015 | A1 |
20220393919 | Knudsen | Dec 2022 | A1 |
Number | Date | Country | |
---|---|---|---|
20220393919 A1 | Dec 2022 | US |
Number | Date | Country | |
---|---|---|---|
62932721 | Nov 2019 | US |