Method of and System for Synchronizing Data Reception and Retransmission Aboard Communication Satellite

Our present invention relates to a method of timing the reading of data bits in a stream of digital signals intermittently received by a relay station, such as a communication satellite, as well as to a system for carrying out this method.

The exchange of digital data between ground stations and a communication satellite may be based on the principle of time-division-multiplex access (TDMA), utilizing several channels of different carrier frequencies each serving for the transmission of messages from a plurality of ground stations in the form of interleaved bursts separated by intervening guard intervals. Aboard the satellite, the data bursts of an incoming message frame are processed for retransmission to their respective destinations in an outgoing message frame of similar structure; the on-board processing includes demodulation of the incoming carrier, regeneration and re-amplification of the base-band signals, and modulation of these signals upon an outgoing carrier. These operations necessarily proceed under the control of a time base aboard the satellite; the bit cadence of the incoming signals, however, is determined by the clocks of their respective ground stations.

Even if the time base of the satellite is controlled by a ground-station clock with the aid of synchronizing signals, its clock frequency will generally deviate from that of the ground station to a greater or lesser extent, e.g. on account of operational delays, Doppler shifts due to relative motion between the satellite and a ground station, and phase differences between the clocks of ground stations accessing the same channel. As a result, distortion or loss of information may occur when the bits received aboard the satellite are read in a rhythm deviating from their rate of arrival.

A possible solution to this problem lies in the use of a buffer memory designed to store a certain number of incoming bits (e.g. three) in respective stages from which they can be read out at a rate different from their rate of loading. This solution, however, necessarily delays the transmission of acknowledgment signals from the satellite to the originating ground stations which serve for the correction of excessive phase shifts between the signal stream emitted by that station and the window allocated to the station in the transmitted message frame. Such a delay, which may amount to the full storage time of the buffer memory, would necessitate a lengthening of the guard intervals between successive bursts in order to prevent interference between messages from different stations; this, in turn, would reduce the capacity of the channel.

The general object of our present invention, therefore, is to provide a method of and a system for more effectively synchronizing data reception and retransmission by a relay station such as a communication satellite, with avoidance of data loss and with maintenance of relatively short guard intervals.

A more particular object is to provide an improved synchronization method and system enabling the use of mutually independent clocks aboard the satellite and at the associated ground stations.

Our invention is based on the availability at the relay station (referred to hereinafter as a satellite) of a clock frequency f.sub.s which is close to the bit cadence in a stream of digital signals intermittently received by the satellite from a remote transmitting station; for reasons that will become apparent hereinafter, the difference between the clock frequency f.sub.s and the bit cadence f.sub.g (determined by the clock frequency of the transmitting ground station) should be so related to the duration of a burst that the drift between clock cycles 1/f.sub.s and bit periods 1/f.sub.g does not exceed one-fourth of a clock cycle for that duration. After extraction of a timing signal corresponding to the bit cadence f.sub.g from a signal stream or burst of limited duration detected aboard the satellite, a square wave of frequency 2f.sub.s generated by an on-board time base is sampled in response to the extracted time signal at an instant substantially coinciding with the midpoint of an initial bit period of the detected signal stream; that initial bit period may be one of several such periods (not necessarily the first one) which carry no message content but merely serve for the detection of the bit cadence, as is well known per se. On the basis of the sample so obtained, i.e. the instantaneous amplitude of the square wave, one of two mutually interleaved trains of synchronizing pulses each having a cadence f.sub.s is selected in such a way that the first selected synchronizing pulse occurs in one of the two middle quarters of a bit period, preferably the second quarter. The subsequent bits of the detected signal stream are then read at the instant of occurrence of the synchronizing pulses of the selected train and are organized into a data frame in step with clock frequency f.sub.s.

According to a more particular feature of our invention, the two interleaved trains of synchronizing pulses are homogeneous pulse flanks (i.e. leading or trailing edges) of the aforementioned square wave and of another square wave of the same frequency 2f.sub.s in phase opposition therewith. The selection may be carried out with the aid of switch means such as a multiplexer having two inputs connected to an output lead of the time base carrying the first-mentioned square wave, the other square wave being obtained from an inverter in series with one of these inputs.

The above and other features of our invention will now be described in detail with reference to the accompanying drawing in which:

FIG. 1 is a block diagram of receiving and retransmission equipment aboard a communication satellite operating in conformity with our invention;

FIG. 2 is a more detailed circuit diagram of a synchronizer according to our invention shown in block form in FIG. 1; and

FIG. 3 is a set of graphs relating to the operation of the synchronizer of FIG. 2.

The equipment shown in FIG. 1 comprises a plurality of receiving antennas AR.sub.1 . . . AR.sub.m each serving a group of incoming message channels whose signals are detected in respective demodulators RD.sub.1a -RD.sub.1n . . . RD.sub.ma -RD.sub.mn. The detected signals from each channel, forming intermittent bursts separated by guard intervals as discussed above, pass into associated signal regenerators RG.sub.1a -RG.sub.1n . . . RG.sub.ma -RG.sub.mn which, as particularly illustrated for regenerator RG.sub.1a, have output leads 1 and 2 extending to respective synchronizers SY.sub.1a -SY.sub.1n . . . SY.sub.ma -SY.sub.mn. Each synchronizer, receiving the detected data bits on lead 1 and a timing signal of frequency f.sub.g on lead 2, has an output lead 3 feeding the incoming data bits, in the rhythm of a local clock frequency f.sub.s generated by a time base CK.sub.s, to a data processor TD for regrouping into outgoing frames, to be modulated upon respective carriers by amplifying modulators MA.sub.1a -MA.sub.1n . . . MA.sub.ma -MA.sub.mn for retransmission via antennas AT.sub.1 . . . AT.sub.m.

Time base CK.sub.s generates a square wave of frequency 2f.sub.s on a lead 6 extending to all the synchronizers SY.sub.1a . . . SY.sub.mn. Under the assumed conditions, each half-cycle of this square wave has a length 1/4f.sub.s approximately corresponding to a quarter of a bit period 1/f.sub.g, thanks to the use of high-stability clock circuits in time base CK.sub.s and at the associated ground stations. Since these clock circuits are independent of one another, there is no correlation between the phases of the incoming bit streams and the on-board clock frequency f.sub.s.

Reference will now be made to FIG. 2 for a detailed description of synchronizer SY.sub.1a which, of course, is representative of all corresponding units in FIG. 1. The synchronizer comprises a conventional discontinuity detector RD connected to lead 2 for receiving the timing signal transmitted over that lead by the associated regenerator RG.sub.1a (FIG. 1) in the presence of an incoming signal stream. Two output leads 4 and 5 of detector RD are de-energized in the absence of this timing signal, i.e. during the guard interval separating successive bursts. Lead 5 is energized at the beginning of the first complete half-cycle of frequency f.sub.g upon the reappearance of the timing signal, i.e. at the midpoint of an initial bit period No. 0 of the data stream on lead 1 as seen in FIG. 3 whose graphs represent the signals on the correspondingly numbered leads of FIG. 2; lead 4 is energized one half-cycle 1/2f.sub.g thereafter, i.e. at the end of the initial bit period No. 0. Although only one bit period No. 0 (shown incomplete) has been illustrated, it should be understood that several such periods devoid of any message content may be provided in order to enable the regenerator RG.sub.1a of FIG. 1 to recognize the bit cadence for the purpose of deriving therefrom the timing signal on lead 2. This timing signal is also applied, via an inverter IN.sub.1, to an input lead 20 of a register RE.sub.I receiving the square wave 2f.sub.s from the output lead 6 of time base CK.sub.s (FIG. 1).

Register RE.sub.I has an output lead 7 extending to a loading input of another register RE.sub.II which has a control input connected to lead 4 and a clearing input connected to lead 5; this register is cleared during the guard interval, i.e. with lead 5 de-energized, and is thus receptive to the amplitude of the square wave of double clock frequency present on lead 6 as sampled at the instant when the timing signal on lead 2 raises the signal level on lead 4 from a logical "0" to a logical "1". The amplitude sample appearing at that instant on lead 7 is stored for the duration of the burst and appears on a lead 8 extending to a switching input of a multiplexer MX. The multiplexer has two data inputs, i.e. an input 6' connected to lead 6 and an input 9 connected to a branch 6" of the same lead via an inverter IN.sub.II ; these two inputs, accordingly, carry two identical square waves of frequency 2f.sub.s which are 180.degree. out of phase. Multiplexer MX has an enabling input 11 which during the guard interval receives a blocking signal from lead 4 by way of a further inverter IN.sub.III also acting as a delay device; thus, the multiplexer is unblocked shortly after the energization of lead 4 to switch its output 10 to one or the other of its inputs 6', 9, depending on the wave amplitude stored at that time in register RE.sub.II. The delay introduced by inverter IN.sub.III amounts to a small fraction of a clock cycle and is designed to stabilize the voltage level on lead 8 before the multiplexer is turned on.

The output lead 10 of multiplexer MX extends to a frequency divider D.sub.I of step-down ratio 2:1 (e.g. a flip-flop of J-K type) which is also disabled during the guard interval by the absence of voltage on lead 5. When the voltage on that lead has changed from logical "0" to logical "1", divider D.sub.I transmits a square wave of clock frequency f.sub.s on a lead 12 to a control input of a register RE.sub.III receiving the incoming data bits from lead 1. Depending on the position of multiplexer MX, the square wave on lead 12 has leading edges coinciding with alternate leading edges of either the original square wave 2f.sub.s on input 6' or the inverted square wave on input 9. These leading edges command the loading of register RE.sub.III with the logical value appearing on an output lead 13 of the register extending to a loading input of a further register RE.sub.IV in cascade therewith. A control input 14 of register RE.sub.IV is connected to lead 6 through another frequency divider D.sub.II similar to divider D.sub.I, input 14 thus carrying a further square wave of clock frequency f.sub.s in fixed phase relationship with the two control waves on multiplexer inputs 6' and 9. The leading edges of the wave present on register input 14 command the readout of the stored bits to the output lead 3 of the synchronizer extending to the data processor TD of FIG. 1.

In the specific instance assumed for the signals shown in FIG. 3, a guard interval terminates at an instant t.sub.0 which marks the start of the initial bit period (or sequence of initial bit periods) No. 0 as shown in graph 1. The timing signal on lead 2, in the form of a square wave of frequency f.sub.g, comes into existence during the first half of this bit period (or of the last bit period of the sequence) and goes to zero at an instant t.sub.1, its trailing edge triggering the detector RD to energize the output lead 5 thereof. At the same time, the square wave 2f.sub.s present on lead 6 is sampled in register RE.sub.I and, as shown in graph 6, is found to have an amplitude of logical level "1" which thereupon remains inscribed on the input lead 7 of register RE.sub.II. An instant t.sub.2, occurring after another half-cycle 1/2f.sub.g of the ground-station clock, the leading edge of the timing signal shown in graph 2 gives rise to the enabling signal on lead 4, causing transmission of the wave sample from lead 7 to the switching input 8 of multiplexer MX. The high signal level on lead 8 switches the multiplexer onto its input 9 which carries the inverted square wave shown in the corresponding graph of FIG. 3; that square wave has a trailing edge which occurs in the second quarter of the first informative bit period No. 1, namely at an instant t.sub.3. It will be evident that the inverted square wave on input 9 would have a trailing edge in the second quarter of this bit period if the wave amplitude on lead 6 had been low instead of high at the sampling instant t.sub.1.

Frequency divider D.sub.I also acts as a phase inverter so that the first leading edge of the wave of clock frequency f.sub.s applied to the control input of register RE.sub.III coincides with the trailing edge of the inverted wave present on leads 9 and 10 at time t.sub.3. At this time, therefore, the incoming data bit appearing on lead 1 in period No. 1 is loaded into register RE.sub.III and read out on its output lead 13. Exactly three quarters of a clock cycle later, at instant t.sub.4, the bit is written into register RE.sub.IV in response to the leading edge of the square wave of clock frequency f.sub.s which is continuously present on the control input 14 of that register; the same bit, therefore, appears on the synchronizer output 3 with a delay of roughly one bit period 1/f.sub.g. That delay would have been foreshortened by a quarter of a clock cycle if the first trailing edge of the timing signal in graph 2 of FIG. 3 had coincided with a low amplitude of the square wave shown in graph 6, with resulting loading of the bit into register RE.sub.III upon the occurrence of a trailing edge of the wave of graph 6 coinciding with a trailing edge of the wave of graph 14.

In an analogous manner, the data bit arriving in period No. 2 is entered first in register RE.sub.III and then in register RE.sub.IV at successive instants t.sub.5 and t.sub.6 coinciding with the next-following leading edges of the waves shown in graphs 12 and 14.

The relative phasing of the waves of frequencies f.sub.g and f.sub.s in graphs 2 and 14 has been exaggerated to show more clearly the drift occurring from one clock cycle to the next. Since the first leading edge of the wave of graph 12 (instant t.sub.3) invariably falls into the second quarter of the first informative bit period No. 1, exactly one reading command loading the register RE.sub.III will be generated during the following bit period as long as the cumulative phase difference does not amount to a drift of more than 1/f.sub.s at the end of the burst.

The incoming data bits, accordingly, are received by data processor TD in step with the local clock frequency f.sub.s, subject only to minor internal delays which can be readily compensated.

Naturally, the inverted square wave fed to the input 9 of multiplexer MX could be generated directly within the time base CK.sub.s rather than in the synchronizer, as shown in FIG. 2, under the control of that time base; the same is true of the square wave of twice the clock frequency f.sub.s supplied to the control input 14 of register RE.sub.IV. It will also be understood that the control signals fed to multiplexer inputs 6' and 9 need not be the illustrated square waves but could be trains of shorter pulses respectively coinciding with the trailing and leading edges of the square wave shown in graph 6 of FIG. 3. These and other modifications, readily apparent to persons skilled in the art, are deemed to come within the scope of our invention as defined in the appended claims.