Decoding algorithm for frequency shift key communications

Abstract

A method and apparatus for decoding binary frequency shift key signals in which an exclusive-OR of the sign of a real waveform with a sign of the imaginary waveform at a time shortly after the real (or, alternatively, the imaginary) waveform crosses zero is used to determine a bit represented by the signal. In some embodiments, particularly those in which the bit period is about one-half of the carrier signal frequency, both the real and imaginary waveforms are monitored to detect the zero crossing in order to account for the situation in which data transitions prevent zero-crossings on one of the waveforms.

Description

BACKGROUND

Track circuits are used in the railroad industry to detect the presence of a train in a block of track. An AC overlay track circuit includes a transmitter and a receiver, with the transmitter configured to transmit an AC signal through the track rails at one end of a block of track and the receiver connected to the rails at the other end of the block and configured to detect the signal. Other than the connection through the track rails, there is typically no connection between the transmitter and receiver for a block. When a train is present in a block of track monitored by a track circuit, the train shunts, or shorts, the two rails, with the result that no signal is received at the receiver. Thus, the receiver uses the presence or absence of a detected signal to indicate whether or not a train is present in the block. It is therefore very important that a receiver in a particular block of interest not interpret spurious signals or stray signals from a transmitter in another block of track or some other transmitter as originating from the transmitter associated with the block of interest.

In order to prevent a spurious signal detected at the receiver from being mistakenly interpreted as originating from the transmitter, transmitters are typically configured to transmit at one of a plurality of fixed frequencies, and personnel responsible for installing the track circuits ensure that all track circuits in close spatial proximity are configured to transmit on different frequencies. In order to provide further assurance that a detected signal originates from a corresponding transmitter, the signal is modulated by a code. In some track circuits, the modulation is performed using a binary frequency shift key technique. With this technique, frequencies above or below the nominal center frequency are transmitted to convey a bit of information (i.e., a frequency above the nominal center frequency represents a logical “1” while a frequency below the nominal center frequency represents a logical “0”). What is needed is a method of reliably detecting these codes. Because there is typically no connection between the transmitter and receiver for a track block, a non-coherent detection method must be used.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic diagram of a receiver and transmitter attached to train tracks according to one embodiment.

FIG. 2 is a block diagram of a tuner/demodulator of the receiver of FIG. 1.

FIG. 3 is a block diagram of a numerically controlled oscillator of the tuner/demodulator of FIG. 2.

FIG. 4 is a block diagram of the filters of the tuner/demodulator of FIG. 2.

FIG. 5 shows plots of a real signal and leading and lagging imaginary signals.

DETAILED DESCRIPTION

In the following detailed description, a plurality of specific details, such as transmission frequencies and track circuit types, are set forth in order to provide a thorough understanding of the preferred embodiments discussed below. The details discussed in connection with the preferred embodiments should not be understood to limit the present inventions. Furthermore, for ease of understanding, certain method steps are delineated as separate steps; however, these steps should not be construed as necessarily distinct nor order dependent in their performance.

FIG. 1 illustrates a track 100 divided into three blocks 0, 1 and 2. The track includes two rails 101, 102. FIG. 1 illustrates insulators 103 at the boundaries of block 1, but it should be understood that the insulators 103 are not necessary and are not present in some embodiments. A transmitter 120 is attached to the rails 101, 102 near one end of block 1, and a receiver 140 is attached to the rails 101, 102 near the opposite end. In practice, the transmitter 120 and the receiver 140 are placed as close to the end of a block as practicable when insulators 103 are present.

FIG. 1 illustrates the receiver 140 for block 1 attached to a transmitter 160 for block 2. As explained more fully in commonly owned co-pending U.S. provisional application Ser. No. 61/226,416, entitled “Track Circuit Communications,” the receiver and transmitter for adjacent block are often collocated because the end of one block is immediately adjacent to the start of another block. FIG. 1 also illustrates a transmitter 120 for block 1, which may be attached to a receiver for block 0 (not shown in FIG. 1).

The transmitter 120 is controlled by a processor 122, which is connected to a memory 126, a serial communications port 128, two 2-bit digital ports 130, 132 and an LED/pushbutton assembly 134. One of the two bit ports 130 is configured for output and the other two bit port 132 is configured for input in this embodiment. As discussed in the aforementioned commonly owned provisional application, the transmitter can accept digital data via the port 128 or 130 to be transmitted via the rails 101, 102. The transmitter may also transmit a fixed code stored in the memory 126. The processor 122 controls a signal generator/modulator 124 to generate a carrier signal at a desired frequency and modulate the carrier signal with digital data or a code using a binary frequency shift key (BFSK) modulation technique. Any conventional BFSK modulator may be employed.

The receiver 140 also includes a processor 142 connected to a memory 146. A tuner/demodulator 144 receives a BFSK signal transmitted via the rails by the transmitter 120 and demodulates the digital data or code carried by the signal. The tuner/demodulator will be discussed in further detail below. Also connected to the processor 142 is a serial communications port 148, two 2-bit digital input/output ports 150, 152 and an LED/pushbutton assembly 154. As discussed in the aforementioned commonly owned provisional application, information demodulated from the received signal may be output to another transmitter (e.g., the transmitter 160) for transmission to a subsequent block, or may be output to an attached wayside device such as a signal.

The tuner/demodulator 144 will now be explained in further detail with reference to the FIG. 2, which illustrates a tuner/demodulator circuit according to one embodiment. Track voltage is baseband sampled via a high speed analog-to-digital (A/D) converter 202 at data rate sufficient to satisfy the Nyquist criterion. In this example, the received signals may have a frequency as high as 20.2 kilohertz and the sample rate is set at 48 kHz. The samples are stored in an 18-bit register (not shown in FIG. 1) in some embodiments. This “raw” track data is first mixed at mixers 204, 206 with the quadrature components (i.e. cosine and sine) of a numerically controlled oscillator (NCO) 208, which is configured to generate a signal at the nominal center frequency expected from the transmitter 120 (which will be noncoherent in some embodiments as discussed above). The mixing of the raw track data with the output of the NCO 208 produces a signal with sum and difference components in a manner well known in the art. For example, if the nominal center frequency is 156 Hz, then the output of the mixer will be a signal that has a frequency component at approximately 312 Hz±2.4 Hz (the sum of the NCO signal of 156 Hz plus the signal detected on the track from the transmitter of 156 Hz±2.4 Hz) and a frequency component of approximately 2.4 Hz (the difference of the NCO signal of 156 Hz and the signal detected on the track from the transmitter of 156 Hz±2.4 Hz).

In some embodiments, the NCO 208 is clocked to produce frequencies ranging from a nominal low frequency of 156 Hz to a nominal high frequency of approximately 20.2 kHz. However, lower frequencies are preferable due to significant attenuation of higher frequency signals in the track rails, with frequencies in the audible range being used most commonly. In some embodiments, 16 distinct nominal frequencies in this range are utilized.

The actual frequency output by the NCO 208 is shifted up or down with respect to the nominal center frequency to represent a logical “1” or a logical “0.” For example, if the nominal center frequency is 156 Hz, a frequency of 158.6 Hz (156 Hz+2.4 Hz) represents a logical “1” whereas a frequency of 153.4 Hz (156 Hz-2.4 Hz) represents a logical “0”. The frequency shift is changed over time to represent individual bits in a multi-bit code. In some embodiments, the codes can be 8 bits long, but longer or shorter code lengths are also possible. In some embodiments, five distinct 8-bit codes (labeled A, C, D, E and F) out of the possible 256 8 bit codes are utilized.

FIG. 3 illustrates the NCO 208 in greater detail. The NCO 208 is preferably of a type sometimes referred to as direct digital synthesis, or DDS. The NCO's output frequency is controlled through an increment register 302, which may be written to by the processor 142 of FIG. 1. The value of the increment register determines how much the phase will change in each clock cycle. The value from the increment register 302 is summed with the previous phase value and input to a phase accumulator 304. The phase accumulator 304 is driven by a clock signal from the clock 305. The output of the phase accumulator 304 is fed back to the phase accumulator 304 and is also summed with the cos/sin select register 306. This value is then used as an index to select a value from the sine ROM 308, which is the output of the NCO 208.

Referring now back to FIG. 2, the mixed signal is then filtered in two stages to remove the sum frequency component such that only the difference component (i.e., the 2.4 Hz component) remains. The first filter 210 is a FIR low pass filter clocked at the ADC sample rate. The first filter 210 acts as an anti-aliasing filter for the second filter with a cutoff (3 dB point) frequency of approximately 20.2 kHz/64=315 Hz. The second filter 212 is clocked by the NCO's address generator to provide 32 times the NCO's programmed frequency f₀ (e.g., where the NCO frequency is 156 Hz, the second filter is clocked at 32×156=4,992 Hz, which is the Nyquist frequency for a 2,496 Hz signal). The second filter 212 is a FIR lowpass filter with a 3 dB frequency at the NCOs' frequency divided by 64 and is designed to reject all frequencies above the difference frequency (f₀/64). The difference in frequency between the cutoff of the first filter (e.g., 315 Hz) and the Nyquist frequency of the second filter (e.g., 2,496 Hz) is sufficient bandwidth. Note the cutoff frequency tracks the programmed received frequency.

FIG. 4 illustrates the first and second filters 210, 212 in further detail to highlight memory allocation, bit width and processing. The coefficients for these filters are shown below in Table 1.

TABLE 1
Filter Coefficients
	Filter 2		Filter 1
	Decimal	Hex	Decimal	Hex
1	488	1E8	−43	FFFFFFD5
2	52	34	−10	FFFFFFF6
3	55	37	−11	FFFFFFF5
4	58	3A	−13	FFFFFFF3
5	61	3D	−14	FFFFFFF2
6	63	3F	−16	FFFFFFF0
7	66	42	−17	FFFFFFEF
8	70	46	−19	FFFFFFED
9	73	49	−21	FFFFFFEB
10	76	4C	−23	FFFFFFE9
11	80	50	−25	FFFFFFE7
12	83	53	−26	FFFFFFE6
13	87	57	−28	FFFFFFE4
14	91	5B	−30	FFFFFFE2
15	93	5D	−32	FFFFFFE0
16	98	62	−34	FFFFFFDE
17	102	66	−36	FFFFFFDC
18	105	69	−38	FFFFFFDA
19	110	6E	−40	FFFFFFD8
20	114	72	−42	FFFFFFD6
21	118	76	−44	FFFFFFD4
22	122	7A	−45	FFFFFFD3
23	126	7E	−47	FFFFFFD1
24	131	83	−49	FFFFFFCF
25	135	87	−50	FFFFFFCE
26	140	8C	−52	FFFFFFCC
27	145	91	−53	FFFFFFCB
28	149	95	−54	FFFFFFCA
29	154	9A	−55	FFFFFFC9
30	159	9F	−55	FFFFFFC9
31	164	A4	−56	FFFFFFC8
32	169	A9	−56	FFFFFFC8
33	174	AE	−56	FFFFFFC8
34	179	B3	−56	FFFFFFC8
35	185	B9	−55	FFFFFFC9
36	190	BE	−54	FFFFFFCA
37	195	C3	−53	FFFFFFCB
38	201	C9	−51	FFFFFFCD
39	206	CE	−49	FFFFFFCF
40	212	D4	−47	FFFFFFD1
41	218	DA	−44	FFFFFFD4
42	223	DF	−40	FFFFFFD8
43	229	E5	−36	FFFFFFDC
44	235	EB	−32	FFFFFFE0
45	241	F1	−27	FFFFFFE5
46	247	F7	−22	FFFFFFEA
47	253	FD	−16	FFFFFFF0
48	259	103	−9	FFFFFFF7
49	265	109	−2	FFFFFFFE
50	271	10F	5	5
51	278	116	13	D
52	284	11C	23	17
53	290	122	32	20
54	297	129	43	2B
55	303	12F	54	36
56	310	136	66	42
57	316	13C	79	4F
58	323	143	92	5C
59	330	14A	106	6A
60	336	150	121	79
61	343	157	137	89
62	350	15E	153	99
63	356	164	170	AA
64	363	16B	188	BC
65	370	172	207	CF
66	377	179	226	E2
67	384	180	246	F6
68	390	186	267	10B
69	397	18D	289	121
70	404	194	311	137
71	411	19B	334	14E
72	418	1A2	358	166
73	425	1A9	383	17F
74	432	1B0	408	198
75	439	1B7	433	1B1
76	446	1BE	460	1CC
77	452	1C4	487	1E7
78	459	1CB	514	202
79	466	1D2	542	21E
80	473	1D9	571	23B
81	480	1E0	600	258
82	487	1E7	630	276
83	494	1EE	659	293
84	500	1F4	690	2B2
85	507	1FB	720	2D0
86	514	202	751	2EF
87	521	209	782	30E
88	527	20F	813	32D
89	534	216	845	34D
90	540	21C	876	36C
91	547	223	908	38C
92	554	22A	939	3AB
93	560	230	971	3CB
94	566	236	1002	3EA
95	573	23D	1034	40A
96	579	243	1065	429
97	585	249	1096	448
98	591	24F	1126	466
99	598	256	1157	485
100	604	25C	1186	4A2
101	610	262	1216	4C0
102	615	267	1245	4DD
103	621	26D	1273	4F9
104	627	273	1301	515
105	633	279	1328	530
106	638	27E	1354	54A
107	644	284	1380	564
108	649	289	1404	57C
109	655	28F	1428	594
110	660	294	1451	5AB
111	665	299	1473	5C1
112	670	29E	1495	5D7
113	675	2A3	1515	5EB
114	680	2A8	1534	5FE
115	684	2AC	1552	610
116	689	2B1	1568	620
117	694	2B6	1584	630
118	698	2BA	1599	63F
119	702	2BE	1612	64C
120	707	2C3	1624	658
121	711	2C7	1634	662
122	715	2CB	1644	66C
123	718	2CE	1652	674
124	722	2D2	1659	67B
125	726	2D6	1664	680
126	729	2D9	1668	684
127	733	2DD	1671	687
128	736	2E0	1672	688
129	739	2E3
130	742	2E6
131	745	2E9
132	747	2EB
133	750	2EE
134	752	2F0
135	755	2F3
136	757	2F5
137	759	2F7
138	761	2F9
139	762	2FA
140	764	2FC
141	765	2FD
142	767	2FF
143	768	300
144	769	301
145	770	302
146	771	303
147	771	303
148	772	304
149	772	304
150	772	304

As discussed above, the output of the second filter 212 is the difference frequency. This output undergoes phase-to-bit processing 214, which correlates the phase of the quadrature signals to a bit (either high or low) and then updates a circular buffer 216 containing the most recent 16 bits received. The phase-to-bit processing may be performed by the processor 142 of FIG. 1, or a separate logic unit which may be realized using a microprocessor, a digital signal processor, a programmable logic unit (e.g., a field programmable gate array), hard-wired logic, or any other method known in the art.

The inventors have observed that the output of the mixers (i.e. the real and imaginary data) shift phase when a frequency shift occurs. In other words, when “high” frequency (representing a “1”) is received, the real signal leads the imaginary signal; but when the low frequency signal (representing a “0”) is received, the real signal lags the imaginary signal. The detection algorithm below takes advantage of this fact by determining whether the real signal is leading or lagging the imaginary signal to determine whether a logic 1 or 0 has been transmitted. This is a fundamentally different approach than that used in BFSK receivers of the type which employ two bandpass filters matched to the high and low frequencies, respectively, to determine whether a logic 0 or 1 has been transmitted.

In some embodiments, particularly those in which one or more full periods of the difference frequency are transmitted before a frequency shift can occur, it is a relatively simple matter to determine whether the real signal leads the imaginary signal. Indeed, a simple and elegant algorithm for doing this is to wait until the real signal changes sign (i.e., when the real waveform crosses the x axis), and then do an exclusive-or between the sign of the real signal and the sign of the imaginary signal after the zero crossing but prior to 90 degrees or one-fourth of the period of the difference frequency. The result will be a logic 0 when the real signal is lagging and a logic 1 when the real signal is leading.

This can be seen with reference to FIG. 5. Shortly after the real waveform crosses zero at zero degrees, its sign is positive (logic 1), and the imaginary waveform is negative (logic 0) when lagging and positive (logic 1) when leading. Shortly after the real waveform crosses zero (i.e., the x axis) at 180 degrees, its sign is negative (logic 0), and the imaginary waveform is positive (logic 1) when lagging and negative (logic 0) when leading. Thus, when the imaginary waveform lags the real waveform, their signs are opposite, whereas when the imaginary waveform leads the real waveform, their signs are the same. The exclusive-or of the sign of the real waveform data and the sign of the imaginary waveform data at a time shortly after a zero crossing (i.e., after the zero crossing but before ninety degrees) therefore indicates whether the imaginary waveform is leading or lagging the real waveform, which also indicates whether a logical 1 or a logical 0 of the code word is being transmitted at the time of the measurement. Those of skill in the art will recognize that the result will be exactly the opposite if the comparisons of the signs of the real and imaginary signals are made at the time that the imaginary signal crosses the x axis.

However, in some embodiments, a single bit of the code is transmitted for only approximately one half period of the difference frequency in order to obtain a high data rate (the period of time corresponding to one bit of the code shall be referred to herein as the symbol period). Moreover, as discussed above, there are frequency drifts between the oscillators in the receiver and transmitter. Due to these facts and the phase shift phenomenon discussed above, it is possible for either the real or imaginary signal to take on the appearance of a full wave rectified sinusoidal signal in which no zero crossing occurs for periods of time in which successive data bits (i.e., symbols) are oscillating between a 1 and a 0. This is because the transmitted frequency is shifting up or down, and therefore the phase is shifting by 180 degrees, on each successive bit, which is a half period of the difference frequency. This phenomenon can take on the appearance of a beat frequency that shifts slowly back and forth between the two waveforms as a function of the frequency drift between the transmitter and receiver oscillators.

However, when no zero crossing on one of the two signals occurs, a zero crossing will necessarily occur for the other signal because of the 90 degree phase difference between them. Thus, the phase-to-bit processing algorithm below selects between the real and imaginary waveforms in order to overcome the problem of no zero crossings on one or the other of those waveforms.

A pseudo code implementation for an exemplary phase to bit processing algorithm is shown below, along with explanatory comments.

TABLE 2
Pseudo-Code Variables
Title	Values	Use	Initialization
Real	0, 1	Sign bit only	0
Imag	0, 1	Sign bit only	0
RealCount	1 thru 1601	Number of samples,	1
		which is proportional to
		time, since a sign
		transition for the real
		waveform (i.e., time since
		a zero crossing of the
		real, or in-phase, signal).
		1600 is between ¾ and
		1 full symbol period,
		which is slightly less than
		one half of the difference
		frequency period.
ImagCount	1 thru 1601	Number of samples,	1
		which is proportional to
		time, since a sign
		transition for the
		imaginary waveform (i.e.,
		time since a zero crossing
		of the imaginary signal).
PhaseRef	−1, 0, 1	reference for edge	0
		detection (i.e. Real or
		Imag)
DataCount	1 thru 2049	2049 is the number of	0
		samples in a data period
BitPointer	1 through	Bit pointer for code	1
	16	register
Data	0, 1	XOR of Real/Imag sign	0
		bits
CodeRegister	16 bit	HW to SW interface	0000 0000 0000
	register		0000
if sign(Real) <> sign(Real[t−1])	// if there has been a zero crossing of the real signal
	Data	= sign(Real) XOR sign(Imag)	// set Data to XOR of the real and imaginary signals
	RealCount = 1	// reset real zero crossing timer
elseif RealCount > 1600 AND RealCount > ImagCount
	// else if no zero crossing of real signal
	// over most of period
	PhaseRef = 1	// set PhaseRef to select imaginary signal data
else
RealCount = RealCount + 1	// otherwise, increment real zero crossing timer
endif
if sign(Imag) <> sign(Imag[t−1])	// if there has been a zero crossing of the imag signal
	Data = NOT [ sign(Real) XOR sign(Imag) ]	// set data to negative of XOR of real and imag signals
	ImagCount = 1	// reset imaginary zero crossing timer
elseif ImagCount > 1600 AND ImagCount > RealCount
	// if no zero crossing of imag signal over most of period
	PhaseRef = −1	// PhaseRef to select real signal data
else
ImagCount = ImagCount + 1	// increment imaginary zero crossing timer
endif
if PhaseRef >= 0
Data = Imag Data	// select imaginary signal data
else
Data = RealData	// select real signal data
end
if ImagCount = 1 AND PhaseRef >= 0	// if the imaginary signal has just been used to
	// determine the value of data bit for this period
	DataCount = 0	//reset the data counter
elseif RealCount = 1 AND PhaseRef <= 0	//else if the real signal has just been used to
	// determine the value of data bit for this period
	DataCount = 0	//reset the data counter
elseif DataCount > 2047	// if the full data period has run
	DataCount = 0	// reset the counter
else
	DataCount = DataCount + 1	//else increment the data counter
endif
if PhaseRef <> 0 AND DataCount = 0	// If either the real or imaginary signal has been
	// selected and one of them has just been used to
	// determine the bit value
	CodeRegister(BitPointer) = Data	// take the data and shift it into circular buffer
	BitPointer = BitPointer + 1	//increment the bit pointer
	if BitPointer > 16	//the code register is 16 bits long
	BitPointer = 1	//reset the bit pointer if at the 16th bit
	endif
endif

In addition to the phase to bit processing described above, the received signals undergo magnitude processing in which the magnitude of the signals is compared to a calibrated reference signal to determine the presence or absence (and, in some cases, distance) of a train. This processing is beyond the scope of the present application and will not be discussed in further detail herein to avoid obscuring the invention.

As disclosed in co-pending U.S. Provisional Application Ser. No. 61/226,416, entitled “Track Circuit Communications” (the entire content of which is hereby incorporated herein by reference), these codes can also be used to transmit information in addition to the codes or during periods when the codes are not being transmitted. For example, if there were 32,768 possible codes (i.e., an 8 bit code word were to be used), then two or more of the distinct code words could be assigned to each transmitter/receiver pair. In the event that two code words were assigned, the transmitter would transmit one of the two distinct codes for a logic “1” and the other of the two codes for a logic “0”. The receiver would be configured to interpret the reception of either of the two codes (or their amplitudes) as evidence of absence of a train in the track block, and would also interpret the received codes as symbols representing bits of data, thereby enabling data transmission using the tracks as the transmission medium. Those of skill in the art will recognize that the baud rate depends on the number of unique codes assigned to a receiver/transmitter pair (i.e., if 4 unique codes are assigned, then each 16 bit code word can represent two data bits, etc.). Alternatively, as discussed above, the transmitter could be configured to transmit a unique code assigned to a receiver/transmitter pair at some periodic rate (e.g., once per minute) and transmit data between the code transmissions. This is possible because the track condition is typically fairly static (and so the need to confirm that received transmissions originate from the correct transmitter is not that time-critical) and because the amplitude of the received transmission of any signal (whether data or code) can be used to detect the presence or absence of a train.

Claims

1. A method for processing a frequency shift key signal comprising the steps of: mixing the frequency shift key signal with in-phase and quadrature components of a signal at a nominal center frequency corresponding to the frequency shift key signal to produce a mixed in-phase component and a mixed quadrature component;filtering mixed in-phase component and the mixed quadrature component to produce a difference in-phase component and a difference quadrature component; andascertaining a logical symbol indicated by the frequency shift key signal at a point in time by determining whether the difference in-phase component leads or lags the difference quadrature component at the point in time, wherein the determining comprises:monitoring both the difference in-phase component and the difference quadrature component to detect a zero crossing of one of the difference in-phase component and the difference quadrature component;performing an exclusive-or operation on a sign of the difference in-phase component and a sign of the difference quadrature component at a point in time that is within the interval between the zero crossing detected in the detecting step and one fourth of a period of the component for which the zero crossing is detected in the detecting step;when the zero crossing is detected for the difference in-phase component, ascertaining the logical symbol directly and solely from the result of the exclusive-or operation on the sign of the difference in-phase component and the sign of the difference quadrature component without calculating a negative of the exclusive-or of the sign of the difference in-phase component and the difference quadrature component; andwhen the zero crossing is detected for the difference quadrature component, calculating a negative of the exclusive-or of the sign of the difference in-phase component and the difference quadrature component and ascertaining the logical symbol directly from the negative of the result of the exclusive-or operation on the sign of the difference in-phase component and the sign of the difference quadrature component.

2. The method of claim 1, wherein the determining step is repeated for a plurality of points in time.

3. The method of claim 1, wherein the component for which the zero crossing is detected is the difference in-phase component.

4. The method of claim 1, wherein the component for which the zero crossing is detected is the difference quadrature component.

5. A system for processing a frequency shift key signal comprising: a first mixer for mixing the frequency shift key signal with an in-phase component of a signal at a nominal center frequency corresponding to the frequency shift key signal to produce a mixed in-phase component;a second mixer for mixing the frequency shift key signal with a quadrature component of a signal at a nominal center frequency corresponding to the frequency shift key signal to produce a mixed quadrature component;a filter circuit operable to input an output from the first mixer and the second mixer, the filter circuit being operable to filter the mixed in-phase component and the mixed quadrature component to produce a difference in-phase component and a difference quadrature component; anda logic unit connected to the filter circuit, the logic unit being configured to ascertain a logical symbol indicated by the frequency shift key signal at a point in time by determining whether the difference in-phase component leads or lags the difference quadrature component at the point in time, wherein the logic unit is configured to perform the determining by: monitoring both the difference in-phase component and the difference quadrature component to detect a zero crossing of one of the difference in-phase component and the difference quadrature component;performing an exclusive-or operation on a sign of the difference in-phase component and a sign of the difference quadrature component at a point in time that is within the interval between the zero crossing detected in the detecting step and one fourth of a period of the component for which the zero crossing is detected in the detecting step;when the zero crossing is detected for the difference in-phase component, ascertaining the logical symbol directly and solely from the result of the exclusive-or operation on the sign of the difference in-phase component and the sign of the difference quadrature component without calculating a negative of the exclusive-or of the sign of the difference in-phase component and the difference quadrature component; andwhen the zero crossing is detected for the difference quadrature component, calculating a negative of the exclusive-or of the sign of the difference in-phase component and the difference quadrature component and ascertaining the logical symbol directly from the negative of the result of the exclusive-or operation on the sign of the difference in-phase component and the sign of the difference quadrature component.

6. The system of claim 5, wherein the logic unit is a microprocessor.

7. The system of claim 5, in which the filter circuit comprises a finite impulse response filter that processes both the mixed in-phase component and the mixed quadrature component.

8. The system of claim 5, further comprising an analog-digital-converter operable to input an analog frequency shift key signal and output digitized samples of the frequency shift key signal to the mixers.

9. The system of claim 8, further comprising a numerically controlled oscillator connected to the mixers and configured to generate a signal at a nominal center frequency corresponding to the frequency shift key signal.

10. The system of claim 5, wherein the determining step is repeated for a plurality of points in time.

11. The system of claim 5, wherein the component for which the zero crossing is detected is the difference in-phase component.

12. The system of claim 5, wherein the component for which the zero crossing is detected is the difference quadrature component.