Decoder and method of decoding for a radiotelephone system employing digitized speech/data and embedded signalling

Background of the Invention

This Invention relates generally to the encoding, transmission, and decoding of digital infoniation in the presence of noise on the transmission channel. More, specifically, this invention relates to a high speed data and digital speech signalling protocol for radio transmission and to a unique system and method of decoding the signalling utilizing a sequence of correlation synchronization words. The coding and decoding techniques are based upon a new coding structure developed herein.

Mobile radiotelephone service has been in use for some time and traditionally has been characterized by a central site transmitting with high power to a limited number of mobile units in a large geographic area. Mobile transmissions, due to their lower power, are received by a network of receivers located remote from the central site and returned to the central site. Due to the limited number of radio channels available, a maximum number of conversations for an entire city would equal the few channels available. Consequently, mobile telephone users discovered that radiotelephone was different than landline telephone due to the often busy conditions of the channels.

To resolve this difficulty, cellular systems were developed to reuse radio channels in a given geographic area. A cellular system characteristically has the coverage area divided into contiguous smaller coverage areas (cells) using low power transmitters and receivers at the central site. One cellular system is further described in U.S. Patent number 3,906,166 assigned to the ) assignee of the present invention. The limited coverage area enables the channel frequencies used in one cell co be reused in another cell geographically separated according to a predetermined plan. One such plan is disclosed in U.S. Patent number 4,128,740, assigned to i the assignee of the present invention. Thus, a large number of channels can be made available in a metropolitan area and the service can appear to be identical co a standard telephone.

The cell system typically utilizes one channel in each cell co receive requests for service (on a reverse set-up frequency) from mobile subscriber units, to call mobile subscriber units (on a forward set-up frequency) and to instruct mobile subscriber units to tune to a frequency pair where a conversation may take place (a voice channel) . The one set-up channel in each cell is continuously assigned the cask of receiving and transmitting data and is the channel to which the subscriber unit tunes when not in a conversational state.

Since the cells may be of relatively small size, the likelihood of a mobile or portable subscriber unit traveling out of one cell and into another is high. To maintain communications, the subscriber unit is handed-off between one cell and another. The cellular systems in use track the unit and decide when a handoff is necessary co maintain quality communications. The subscriber unit i.s commanded, via a high speed data message interrupting the audio communications on the voice channel, to retune the transceiver to another frequency which is available in a new cell. This handoff 5 requires a relatively short period of time and the user is generally unaware of the occurrence.

Since cellular telephone systems provide performance characteristic of the land line telephone system and interconnect with it, subscribers expect land telephone system features from the cellular telephone system. One such feature is the transmission of data from one location to another. Many telephone subscribers connect data communications devices, such as a personal computer, to the telephone system via a modem. Modems are familiar 15 to those skilled in the art and fundamentally operate by converting data 1 and 0 levels to distinct tones or to particular tone waveform phase relationships which can be transmitted by the land telephone network.

It would be natural to connect a computing device 20 via a modem to a radiotelephone subscriber unit for communication with another data generating device via the land telephone network. In fact, this has been done and produced unsatisfactory results. Rapid multipath fading, commonly experienced in high frequency cellular radiotelephone communications, causes gaps and significant phase changes in modem-generated tones such that data carried by the radio channel becomes garbled or missing. Furthermore, a handoff between cells, which to a human engaged in conversation is virtually unnoticeable, becomes a formidable obstacle for the communication of data generated by a data generating device.

This limitation has been resolved by converting the input data to a data format compatible with radio channel 35 transmission and reconverting the radio channel data format back to the original format upon reception. The data transmission is halted prior to a handoff and is resumed after the handoff is completed. A detailed description of the above process may be found in U.S.

Patent No. 4,654,867 and assigned to the assignee of the present invention.

A second feature of the landline telephone system which subscribers to a cellular radiotelephone system wish to have is that of relative security of their conversations. Digital scrambling techniques for secure radio communication channels has been disclosed in U.S. Patent nos. 4,167,700; 4,434,323; and 4,440,976 each assigned to the assignee of the present invention. These inventions, however, do not address the aforementioned handoff requirement of cellular systems and do not provide the necessary transponding of a supervisory signal to maintain communications on an assigned channel.

Cellular systems, to reduce interference between cochannel users, employs several distinguishable signals (SAT signals) to identify cellular fixed sites. Each site is assigned one signal which is transmitted with each transmission from fixed site to subscriber unit. The subscriber unit, in turn transponds the same signal to the fixed site and the radio channel connection is to the fixed site and the radio channel connection is maintained. If anomalous propagation enables cochannel transmission to or from one cell to be received in another cell, the lack of a properly transponded signal will cause the radio channel to be deallocated from the interferring subscriber unit. These distinguishable SAT signal typically are unique tone frequencies for analog transmissions and unique bit sequences for digital transmissions.

Th. present invention serves the requirements of radiotelephone subscribers by providing at least three modes of operation which are automatically selected and verified by the system via the signalling information transmitted between fixed site and subscriber unit, λ common data format is employed for both data transmission and digital voice privacy and continuous bit synchronization may be maintained from subscriber unit through the fixed site to the landline instrument.

Conventional coding and decoding methods do not take full advantage of the latent error detection and correction power of a coding technique employing a sequence of multiple bit synchronization words to carry information. Error detection and correction, as previously practiced, relies upon redundancy of information, correlation techniques, or parity to correct one or more bits in an information bit string. Each bit in the information string is either correct or in error. The present invention utilizes a third possible state in the information bit string - a missed bit. Each bit in the system control message of the present invention is derived from a synchronization bit sequence word, itself consisting of a plurality of bits.

Summary of the Invention

Accordingly, it is an object of the present invention to include synchronization, supervisory (SAT) information, and system state in a fixed bit length synchronization word.

It is a further object of the present invention to enable clear analog transmissions, digital privacy speech transmissions, and data transmissions to occur between fixed site and subscriber units in a cellular radiotelephone system.

It is a further object of the present invention to employ high autocorrelation, low cross correlation multiple bit words and their ones complement inverses as a set of synchronization words.

It is a further object of the present invention to code system state messages into predetermined sequences of synchronization words.

It is a further object of the present invention to receive and decode the system state messages from a noisy channel with error detection and correction enhanced by the capability of detecting a missed eynchronization word and substituting a correct synchronization word in place of the missed word.

It is a further object of the present invention to employ a common data format for both digital privacy and data transmissions.

It is a further object of the present Invention to provide continuous bit synchronization from subscriber unit through the fixed site to the landline privacy or data terminal.

According to the invention there is provided a decoder for decoding system states conveyed in sequences of multi-bit normal words and ones complement inverse words used for synchronizing digital messages transmitted over a radio channel which may introduce bit errors into said words, said decoder characterized by:

means, responsive to the introduced bit errors, for detecting a normal word, an inverse word, and a word miss and for assigning a first binary level to said detected normal word, a second binary level to said detected inverse word, a randomly selected one of said binary levels based on best-guess information to said word miss, thereby creating a selected sequence of binary levels;

means for generating a decoded sequence of binary levels corresponding to one of the system states; and means for comparing said selected sequence and said decoded sequence and for indicating a correct system state decode when said selected decoded sequences have a predetermined correlation therebetween.

Brief Description of the Drawings

Figure 1 shows a block representation of a conventional three cell cellular system.

Figure 2 is a timing diagram of the signalling format used in the forward set-up channel of a conventional cellular system.

Figure 3 is a timing diagram of the signalling format used on the reverse set-up channel of a conventional cellular system.

Figure 4 is a timing diagram of the signalling format used on the forward voice channel of a conventional cellular system.

Figure 5 is a block diagram of the subscriber unit and the fixed site equipment which may be used to realize the present invention.

Figure 6 is a schematic diagram of a Manchester data encoder which may be utilized in the present invention.

Figure 7 is a block diagram of a digital cellular signalling control (DCSC) circuit which may be employed in the present invention.

Figure 8 is a timing diagram of the generalized synchronization format employed in the present invention.

Figure 9 is a timing diagram of the synchronization and digitized speech mode format employed in the present invention.

Figure 10 is a tining diagram of the data halt transmission beginning the digitized speech mode employed in the present Invention.

Figure 11 is a timing diagram of the data format for 5 digital signalling tone (DST) employed in the present invention.

Figure 12 is a timing diagram of the data format far 1200 bits per second data transfer employed in the data terminal mode of the present invention.

Figure 13 is a timing diagram of the data halt format employed in the present invention.

Figure 14 is a timing diagram of the change between data terminal mode and data hold mode of the present invention.

Figure 15 is a timing diagram of the digitized speech (RF only) mode employed in the present invention.

Figure 16 is a flowchart of the DCSC synchronization (S/S) word soft detect subroutine.

Figure 17 is a flowchart of the DCSC S/S word mode 20 detect subroutine.

Figure 18 is a flowchart of the DCSC S/S word synchronization maintenance subroutine.

Figure 19 is a flowchart of the DCSC S/S word reacquisition subroutine.

Figure 20 is a flowchart of the DCSC S/S word detection background subroutine.

Figure 21 is a flowchart of the subscriber unit DCSC radio scan background subroutine.

Figure 22 is a flowchart of the DCSC S/S word 30 transmission subroutine.

Figure 23 is a flowchart of the transmit subroutine employed in Figure 22.

Figure 24 is a flowchart of the DCSC digital signalling tone (DST) transmission subroutine.

Figure 25 is a block diagram of the synchronization word detection correlator employed in the present invention.

Figure 26 is a block diagram and output table of the

S/S pattern detector employed in the present invention.

Figure 27 is a three dimensional vector diagram in which the diagram comers represent possible synchronization code sequences and the circled comers illustrate a code of Hamming distance two.

Figure 28 is a timing diagram Illustrating the three possible sequence bock S-S-Sן S-S-S; and S-S-S available when one synchronization word is transmitted as inverse.

Figure 29 is a three dimensional vector diagram illustrating that two synchronization word misses can be corrected when the code is of Hamming distance three.

Figure 30 is a block diagram of a synchronization sequence decoder which may be employed in the present invention and which utilizes a ROM table to determine the closest synchronization word sequence.

Figure 31 is a block diagram of a synchronization * sequence decoder which may be employed in the present invention and which is capable of correction of all synchronization misses up to the capability of the code and detection of synchronization errors.

Figure 32 is flowchart for a synchronization sequence decoder which may be employed in the present invention and which is capable of correcting combinations of synchronization errors and synchronization misses.

Figure 33 is a block diagram of a miss counter for logic distance 2.

Detailed Description of the Prefarred Embodiment λ representation of a multisite radio system such as that which nay be employed by a cellular radiotelephone system is shown in Figure 1. In this representation, a geographic area is shown divided into three of a possible 5 many radio coverage areas (102, 104, and 106) which are commonly called calls. Generally located within each cell is a configuration of fixed site equipment shown as fixed site equipment 108, 110, and 112. The fixed site equipment typically consists of receivers, transmitters, and a site controller. In one implementation of a cellular system, the fixed transmitters and receivers are located at the center of each cell and illuminate this cell with radio signals omni-directionally or directionally. Other cellular implementations place the fixed site at the periphery or elsewhere within the cell. Further discussion of a center illuminated sectorized cellular system may be found in U.S. Pacent no. 4,128,740 assigned to the assignee of the present invention.

A plurality of subscriber units are typically present in each cell and are represented in Fig. 1 by subscriber units 114, 116 and 118. Any particular subscriber unit may be vehicle mounted or may be hand carried. Each subscriber unit in a cellular radiotelephone system has the capability of initiating and maintaining a telephone call via one of the transmitters and one of the receivers of the fixed site equipment and a cellular telephone exchange 120. The cellular telephone exchange 120 performs call routing and public switch telephone network interface and may be a device such as an EMX 500 marketed by Motorola, Inc. The cellular telephone exchange 120 may accept a call from either the public switch telephone network or a subscriber unit and route that call to its proper destination.

In a conventional cellular system, one duplex radio channel is assigned the task of transmitting information such as call requests, voice channel assignments, handoff instructions as a served subscriber unit travels out of the radio coverage area of one cell and into the radio coverage area of another cell, and system maintenance instructions. On the inward or reverse set up channel half of the duplex channel, the fixed sice controller receives requests for service and other functions from the subscriber unit. Actual message transmission occurs on another duplex channel, commonly known as a voice channel, available within a radio coverage cell.

Thus, a telephone call may come in from the switched telephone network to the cellular telephone exchange 120, be routed through fixed site equipment 110 which determines if a voice channel is available in cell 104 and instructs subscriber unit 116 co tune to che unoccupied voice channel via the forward set up channel.

The subscriber unit 116 tunes to the designated voice channel and message conversation may begin.

To provide a check of the continuing operation of 5 the subscriber unit during a conventional cellular radiotelephone call, a tone, equivalent to the DC supervision in normal telephone operation, is transmitted continuously from the fixed site controller via the fixed transmitter, received by the subscriber unit and retransmitted by the subscriber to the fixed site receiver. This tone is known as supervisory audio tone (SAT), and is used inter alia, to control cochannel interference. In conventional cellular systems, small offsets in the frequency of the SAT are used to identify 15 each cell and if the subscriber unit having service does not transpond the proper SAT, the call will be terminated.

In a cellular system having relatively large numbers of subscriber units, a large amount of information must 20 be transmitted and received on the set-up channel.

Efficient design suggests that the data be organized into a synchronous format of fixed length words and synchronizing pulses. The conventional format for signalling on a cellular system forword set-up channel is 25 shown in Fig. 2 and the format for the cellular system reverse set-up channel is shown in Fig. 3. One source of poor reliability is error introduced in the radio channel by Rayleigh fades caused by subscriber unit motion through the multi-path interference patterns generated by 30 reflection from obstacles near the receiving equipment.

These errors generally occur in dense burrs with the average burst duration related to the average fade length. Additionally, the bit error probability is substantially independent of data rate until bit lengths 35 approach che average fade duration. This characteristic indicates chat to minimize errors, data rates should either be very low or as high as channel bandwidth allows. Due to the amount of information to be transmitted and the availability of conventional error correction techniques, a 10 KBS data rate was chosen for the standard data rate of the cellular system in the United States.

System control data messages may also be sent on the voice channels. These messages are primarily handoff messages directing a subscriber unit to tune to another channel but may also include other control functions such as subscriber unit transmitted power control. In a conventional cellular system the technique used is blank and burst in which the voice signal is muted and a data signal, shown in Figure 4, is sent at a.10 kilobit rate.

The standard landline telephone employs a DC current to indicate whether the telephone user׳ is on hook or off hook. Host cellular radio telephone systems employ a tone analogous to the DC ־current for call supervision. By employing slightly different frequencies of analog tones, co-channel interference between neighboring cell radio coverage areas may be reduced. The tone is transmitted from the fixed site equipment on an active voice channel and is received and retransmitted by the subscriber unit using the channel. If another tone frequency is returned, the fixed site controller interprets the incoming signal as being corrupted by interference and the audio path is muted.

As users travel through the cellular system they may reach the coverage limit of one cell and therefore be better served by another cell. The process of switching the call from one cell to the other is known as handoff. When a need to handoff a subscriber unit from one cell co another occurs, a message is sent to the subscriber unit in the blank and burst format on the voice channel. The subscriber unit mutes the audio and tunes to a radio channel which was indicated in the olank and burst data message. The audio is unmuted when the subscriber unit has tuned to the new channel.

It has become increasingly desirable to provide cellular radio telephone systems with the capacity to carry more than analog voice messages. Purely data messages, such as those generated by data terminals or by digitized voice devices, has become increasingly common. One method of conveying purely data messages via a cellular radio telephone system was shown and described in U.S. Patent No. 4,654,867, Labedz et al. and assigned to the assignee of Che present invention. There, data messages were encoded into a format compatible with the prior art data formats previously described. Additionally, methods for preventing data loss during handoff and for providing data signals equivalent to the supervisory tone were shown and described. The invention of the present application provides a means for encoding system control information for high speed data or digitized voice, transmitting the same over a cellular radiotelephone system, and decoding the system control information with, minimized errors.

The two basic components of the system of the present invention are shown in Figure 5. A block diagram of a subscriber unit 116 is shown communicating with a fixed site 110. A signal to be transmitted by the subscriber unit 116 is input to speech/data converter 501 which performs an analog to digital conversion and may oe implemented with an MCI4402 Codec available from Motorola, Inc. A digital signal, representative of the input speech or data, is coupled to a digital cellular signalling control (DCSC) function 503 via a 9.6 KBS transmit data line 505 and a 9.6 KBS transmit clock line 507. Additionally, a 10 KBS clock line is coupled between the speech/data converter 501 and the DCSC 503 on line 509. The unique functions perforaed by the DCSC 503 will be described later. Binary data at a rate of 10 KBS is output from DCSC 503 to conventional Manchester encoder 511 via 10 KBS transmit data line 513 and 10 KHz clock line 515. A Manchester encoder 511 may be implemented with discrete logic functions as shown in Figure 6. Ten KBS transmit data may be input on line 513 to a conventional DQ flip-flop 601 which is clocked by the 10 KHz clock input on line 515. The Q output of flip-flop 601 represents a clocked delay of the input and is input to exclusive OR 603 where it is exclusive OR-ed against the 10 KHz clock to produce an output which is a Manchester encoded data corrupted by transients. This output is coupled to DQ flip-flop 605 which is clocked by a 20 KHz clock signal conventionally derived and having edges coincident with the 10 KHz input signal to remove .'the undesirable transients. The Q output of flip-flop 605, on line 607, is a conventional Manchester encoded version of the input data.

Returning to Figure 5, the output of Manchester encoder 511 is input to the transmitter of transmitter/ receiver 517 which transmits the signal on a channel assigned by the cellular system. The transmitter/ receiver 517 may be any mobile or portable transceiver compatible with a cellular system and is generally described in service manual 68P81066E40-0 available from Motorola, Inc.

The transmitted signal is received by a fixed site receiver of fixed transmitter/receiver 519 which is part of the fixed site equipment 110. The fixed transmitter/ receiver 519 may be any fixed radio equipment suitable for use with cellular telephone systems and is further described in service manual 6bP81060E30-0 available from Motorola, Inc. The data signal is demodulated from the radio carrier by the received and coupled to a signalling interface circuit 521 which is part of the site controller 522 and which recovers 10 KBS binary data and a 10 KHz clock for presentation to the fixed site DCSC

523 via lines 525 and 527 respectively. ‘A VCP 528 (Voice

Control Processor such as that described in the Fixed

Network Equipment service manual no. 68P81052E50 and available from Motorola, Inc.) performs the functions of controlling the receive and transmit baseband audio, receiving SAT, and relaying SAT and other system control messages to higher level system processors (not shown). The DCSC 523 is similar to DCSC 503 and will be described later. The 9.6 KBS recovered data and its clock is presented to a conventional modem 529 via lines 531 and 533, respectively, for transmission down conventional wire line facilities to the cellular telephone exchange. Digital speech/data from the cellular telephone exchange is input to modem 529 for conversion to 9.6 KBS data and clock and coupled to DCSC 523 via lines. 535 and 537 respectively. The modem 529 also produces a 10 KBS clock synchronous with the 9.6 KBS clock for presentation to DCSC 523 via line 538 and subsequent coupling to

Manchester encoder 539 (which may be an implementation identical to Manchester encoder 511) via lines 541 and 543 respectively. The Manchester encoded signal is coupled to the fixed transmitter/receiver 519 for transmission to the subscriber unit 116.

The subscriber unit 116 receives the transmitted signal and demodulates it in the receiver of transmitter/receiver 517 and couples the demodulated Manchester encoded 10 KBS signal to signalling interface circuit 545. The signalling interface circuit 545 recovers the 10 KBS binary data and clock and presents them co DCSC 503 via lines 547 and 549. The transmitter receiver 517 also provides a two bit digital representation of the digital message equivalent co the supervisory tone employed in conventional cellular telephone systems on line 551. The DCSC 503 provides a detect Logic level to transmitter/receiver 517 on line

553 to indicate presence of the proper digital SAT signal. The DCSC 503 couples 9.6 KBS data and clock to the speech/data converter 501 on lines 557 and 559 respectively. Speech/data converter 501 regenerates the 5 original analog or data signal input from the cellular telephone exchange.

The digital cellular signalling control (DCSC) function 503 is shown in detail in Fig. 7. The block diagram for DCSC 523 is similar except that an additional 10 output of digital signalling tone detect is output to the VCP 528 on line 561. SAT detect is coupled to VCP 528 via line 563. The digital cellular signalling control (DCSC) is a microprocessor based data conversion and interpretation unit which employs a microprocessor 701 15 (which in an preferred embodiment may be an MC 6809 marketed by Motorola, Inc.) and associated memory. The memory is realized by conventional read-only memory (ROM) 703 and random access memory (RA:;) 705. A timer 707 times the position and duration of SAT/SYNC words. A 20 peripheral interface adapter (PlA) 709 provides an interface function between the microprocessor 701 and the other devices on the bus and performs the functions of bus buffer and latch. SAT/SYNC correlator 711 and SAT/SYNC correlator 713 are conventional bit correlators and are described in conjunction with Figure 25. Address lines A-13 through A-15 are coupled from microprocessor 701 to address decoder 715 and are used as chip select Lines for SAT/SYNC correlators 711 and

713, timer 707, ROM 703, RAM 705, and PIA 709.

The unique format for transmitting digitized speech, terminal data, and special control messages over a cellular voice channel employed by the present invention is shown in Fig. 8. With a 10 KBS channel bit rate, this 35 format effectively provides a high speed 9.6 KBS data race over a conventional cellular radio telephone system channel bandwidth.

The SAT/SYNC (S/S) word is a 21 bic correlator word which uniquely provides combined radio frequency (RF) frame synchronization, digital SAT (supervisory tone) information, and system state information. A sec of six 21 bic correlator words have been developed for che SAT/SYNC function in the preferred embodiment. This set consists of three normal correlator words (S/S)!, (S/S)2 and (S/S)3 plus their ones complement inverses (5/S)!, (S/S)2, and (S/S)3. This set may be designed co provide high auto/correlation properties (i.e. a 21 bit match when the word is aligned in the correlator and less than or equal to 2 bit match when the word is not aligned) and low cross correlation properties (a 21 bit match when aligned and a less than or equal to 6 bit match when not aligned) within the.set by those skilled in the art. The method by which these auto correlator words are used to provide SAT information, speech and data RF frame synchronization, and system information is as follows:

Each of the three 21 bit S/S correlator words and its inverse corresponds to one of the three SAT frequencies in general use by cellular radiotelephone systems (5970, 6000, and 6030 Hz). During a cellular call, one normal S/S and/or its inverse is used to convey supervisory information from the fixec site to che subscriber unic. The subscriber unit must then detect the S/S or ics inverse and transpond it to the fixed site. The 21 bic correlators (711 and 713 of Fig. 7) ac che subscriber unit and ac che fixed sice controller are programmed co detect one of three SAT words and ics inverse. Due to che low cross correlation between words, che probability of one SAT falsing another is very small.

The low noise and cross SAT falsing probabilities are made possible by using all 21 bits for both correlation and for synchronization, rather than splitting the overhead bits for separate signalling tasks.

SAT information is conveyed between the fixed site 5 and the subscriber unit by transmitting either S/S or its inverse 5/S. This allows system mode information to be continuously transmitted between the subscriber unit and the fixed site by encoding the sequence of transmitted S/S and its inverted words. System mode 10 changes are communicated between the subscriber unit and the fixed site with automatic confirmation of the mode made via the transponded SAT signalling. SAT signalling is never interrupted during mode changes by utilizing the full 21 bit S/S correlator word for system mode definition. Since 21 bits are used in each S/S word detection, a high degree of protection is provided against incorrect mode changes. Additionally, this method also allows a double frame format for transmitting data terminal messages which will be described later.

The format for transmitting digitized speech at 9.6 KBS is shown in Fig. 9. This is a more specific format than that shown in Fig. 8 in that the 504 bits of data comprise digitized speech. The ratio of 504 bits of digitized speech to 21 bits of overhead (SAT/SYNC) information provides an effective speech throughput of

9.6 KBS. SAT is continuously sent by transmitting the proper correlator word, S/S, which־corresponds to the correct SAT frequency.

In the preferred embodiment, digitized speech mode 30 is defined by transmitting only inverse S/S words for RF frame sync and SAT information. Four S/S words and their cyclic permutations define the system mode, (digital speech « 0000; digital speech (RF only) 1111 «־; digital data 1010 «־ or 0101; and digital halt 1 001 ,0110 ,1100 .־ , 35 or 1001). In operation, a cellular radiotelephone call may be originated in the clear (non-digital) mode where the conventional analog SAT is sent and transponded. When the radiotelephone user wishes to switch to che digital speech mode, as he might if he were using a digital voice privacy device sometimes colloquially known as a scrambler, a reverse voice channel message is sent to the fixed site requesting a digital channel for private speech mode. After the fixed sice receives the digital request message, it engages che digital site controller 522 of Fig. 5 ana transmits the digital SAT format shown in Fig. 10 on the assigned radio channel. The 504 bits of dotting״ information (which may be the bit pattern 001100110011..) is used for bit synchronization. This subscriber unit then transponds the digital SAT signalling format to the-fixed site, confirming that the subscriber unit is in the correct mode and is on the correct digital channel. Once in the digital mode, all further mode changes and confirmations are made by changing the S/S and che ־S/S sequences.

If a digital channel to be assigned to this subscriber unit is not the same as the channel currently used by this subscriber unit, a handoff command is transmitted from the fixed site to the subscriber unit to switch to a channel capable of digital service. The subscriber unit repeats the aigital service request message until it receives the correct digital SAT and speech mode format.

To set up a digital privacy call (for example to transport public key information), the 504 bit dotting pattern may be replaced with encryption set-up data. After the digital privacy call is set-up, the 504 bit frame is used to transmit che digitally encoded speech at an effective rate of 9.6 KBS. The digitized speech data >.

is subsequently used to derive bit synchronization between the S/S correlator words. If a predetermined number of consecutive S/S sequences are lost, speech is muted. When digital voice privacy is employed, continuous bit synchronization from the subscriber unit through the fixed site to a landline encryption device allows proper operation of the encryption key generator 5 without requiring periodic key generator data transfer. In the preferred embodiment, the bit synchronization withstands RF phase jumps, channel fading, drift between RF and landline modem clocks, and bit slippage due to an out-of-lock phase locked loop at the subscriber unit or 10 the fixed site.

Digital signalling tone information can be transmitted coherently with SAT, using the format shown in Fig. 11. Twenty-four inverted (3/S) words of the same SAT frequency are inserted in the 504 bit 15 frames. The 21 bit S/S words continue co be used to define SAT information, system mode, and frame timing. In the preferred embodiment, the fixed site may have a flexible algorithm to detect a majority number of the 24 inverted (S/S) words transmitted every frame of

525 bits. This method provides a very high signalling tone detect probability, while keeping the falsing probability low.

Transmission of data at a rate of 1200 BPS computable with data terminals is shown in Fig. 12. 25 Alternate S/S and S/S synchronization words indicate that the system is in the data terminal mode. SAT information is conveyed, as described previously, by the selection of which S/S and S/S words are transmitted. Inverse synchronization words, S/S, 30 may also be used to define double length frames, 1050 bits, which permit the utilization of forward control channel error correction. In Che preferred embodiment, this error correction uses 5 word interleaving (A, B, C, D, E) for improved burst error protection at lower 35 vehicle speeds.

Figure 12 illustrates that two 504 bit frames may be defined by alternating S/S and 3/3 words.; If frame synchronization is lost, this format allows quick double frame synchronization recovery. 500 bits from each frame are concatenated to form a 1,000 bit frame. This frame allows five repeats of 200 bit segments; each segment forming five different 40 bit words (A, B, C, D, E). At the receiver, the five 200 bit segment repeats are majority voted. The resulting five words of 40 bit each are then forward error corrected using (40, 28) BCH error correction. This technique results in 140 bits of corrected data every two frames (105 milliseconds). Thus, an effective data throughput of 1333.3 BPS, more than the required 1200 BPS rate, is provided. The added throughput can be used for modem control- information, such as block parity and number of characters per block. Alternative error correction techniques can use the 1008 bit double frame data block to provide a higher effective data rate. A 2.4 KBS throughput requires 252 bits of data per double frame, which results in exactly a 1/4th rate error correction.

Automatic digital data mode capability is provided by detecting an alternating sequence of S/S and 3/S words. Due to the low cross correlation property between an S/S correlator words and its inverse (as chosen in the set of six correlator words), plus utilization of the entire 21 bit S/S word, a high degree of cross falsing protection is provided in system.־ mode determination.

After the conversation begins, all cellular radiotelephone system control messages (such as handoff or power change) are transmitted over the voice channels. While digitized speech can generally tolerate a message overriding the speech, data terminal transmission must be halted. Figure 13 shows the preferred format which implements the data halt mode. As the figure indicates, an inverse SAT/SYNC word is included at every second word. This mode can be invoked by either the fixed site or the subscriber unit. While in the data halt mode, special control messages (CONA, CONB) can be transmitted between the fixed site and the subscriber unit. For example, parity and data rate messages can be sent from the subscriber unit to the fixed site modem before the data transmission mode is initiated or after a handoff to set up the new fixed site modem.

Utilization of the data halt format as shown in Fig. 1 3 also provides a double frame synchronization for utilizing the same error control used during the transmission of 1200 baud data in the data terminal mode. CONA and CONB form a concatenated 1000 bit data frame, synchronized by the S/S words as shown in Fig. 13. When changing from data terminal transmit mode to data halt mode, the S/S word sequence as shown in Fig. 14 is used. This sequence changes modes without losing double frame synchronization continuity or SAT information. This data format may be used to send special messages (CONA, CONB as shown) or may temporarily continue to send data terminal information.

For example, to transmit a handoff command during the data terminal mode, the fixed site will switch to the 25 data halt mode shown in Fig. 13 prior to sending the handoff control message. Data terminal information continues to be transmitted within the data halt moae format using the double frame synchronization. The subscriber unit detects the data halt mode, stops mobile 30 to landline data transmission, and sends a message to the landline terminal to halt data transmission. Once the subscriber unit determines that landline to mobile data transmission has been halted, the subscriber unit transponds the new halt mode with its SAT information to the fixed site. The fixed site may then detect that data is halted in both direction and a handoff can proceed.

After the handoff, the destination fixed site initially transmits SAT in the data halt system mode. The subscriber unit can send special messages to the fixed site in this mode to set-up a modem in the new cell.

After the modem has been set-up, the fixed site may return to the transmit data terminal mode. Digital SAT transmission is never interrupted during these system mode changes.

A special digital speech mode can be used to provide 10 speech privacy between the subscriber unit and the fixed site while providing clear speech to the cellular telephone exchange and the switch telephone network. The path most susceptible to interception, then, is protected without requiring the land end user to use a scrambling 15 device. The signalling for this mode is shown in Fig.

15. Additionally, digital signalling tone information can be transmitted־ coherently with SAT during this mode using a format similar to that shown previously in Fig.

except that the SAT/SYNC words are inverse.

Th. method of detecting SAT/SYNC signalling employed by the digital cellular signalling control (503 or 523 in Fig. 5) is illustrated in ths flowcharts 0f Figs. 16, 17 and 18. Referring first to Fig. 16, the DCSC is first reset at 1601 and the hardware is initialized at 1603 upon equipment power up. As has previously been described, the SAT/SYNC consists of 21 bit words transmitted at predetermined locations during data transmission, system mods information is contained in the transmission of particular sequences of the three S/S words and their inverses. Detection of the particular sequence is achieved by correlation of a received S/S word to a predetermined stored word or inverse and subsequent address determination from a binary representation of the S/S word sequence. Further description is given in conjunction with Figure 26. The first series of steps in the signalling routine performs the function of a soft detection which-in essence is the prevention of falsing. Essentially the received data is correlated with the expected one of the three S/S words and its inverse in a conventional fashion. The expected word is provided by the binary representation coupled to the DCSC on lines 551. A detection of a received S/S word must occur for the system to proceed. Following the first word detection, one more detection of an S/S word must occur at the proper time within the next three frames, allowing one bit error. This is accomplished as follows: a bit error threshold is set to zero allowable errors at 1605 following initialization 1603. A window of time, during which the received data is examined for correlation, is opened at 1607. during the time window the data is tested for data correlation. When an errorless correlation is detected at 1609 the timing of the next window for the next S/S word is set and the first window time is closed at 1611.

Following a first S/S word detection, the error thresholds is set at one allowable bit error at 1613 and detect registers for the S/S word are updated at 1615. At a time determined by the set next window step 1611, a second timing window is opened at 1619 and closed a predetermined time later at 1621. An interrupt driven background subroutine, which will be described later, determines whether a correlation has occurred within the proper time window. The detect registers are subsequently updated at 1623 and a test for the second S/S word detect is made at 1625. If the second S/S word is detected, the program moves to the next portion of its operation at A. If a second S/S detect is not found at 1625, a determination of whether three consecutive S/S word detections have been missed during the proper window time at 1627. IF three consecutive misses have net occurred, the program returns to the window timing sequence at 1619. If three consecutive misses have occurred, the program returns to the first word soft detect sequence starting at 1605.

Referring now to Fig. 17, as a result of the previously described steps, digital SAT has been detected and the digital SAT output of the DCSC is set high at 1701. The program then moves to the system mode detection sequence and starts by setting the bit error threshold to two at 1703. Window timing, which is established by previous S/S detects, is opened at 1705. During the time window, determination of the presence of digital signalling tone (DST) detection for the previous frame is made at step 1707 and the DST output is set high or low depending upon the detected presence or absence of digital signalling tone respectively. (DST presence is determined at this time window because DST is not present during the S/S word, therefore the determination does not interfer with DST detection operations). The window is closed at the proper tine at step 1709 and the detect registers are updated at 1711. The system node is determined at 1713 iron the detect registers and a successful decode causes ths system node to be output iron the DCSC at 1715. An unsuccessful systen mode decoding at 1717 causes the subroutine to loop back to the synchronization window opening at 1705. Four consecutive S/S word misses at 1719 returns the process to the Initial SAT detect subroutine at 1505 after setting the digital SAT output low at 1721.

Referring now to Fig. 18, an ongoing detection of synchronization and systen mode is accomplished by a continuing check of S/S word correlation (allowing for three consecutive S/S misses at 1801) and system mode decode at 1803. If four consecutive S/S word misses occur, it is assumed that the received synchronization is lost and a reacquisition sequence is attempted at shown ' in Fig. 19 following the setting of the digital SAT output low at 1805.

Reacquisition is attempted by setting the bit error threshold to zero at 1901 and setting the S/S window to the expected time plus or minus a larger predetermined range than before, which in the preferred embodiment is plus ar minus 600 microseconds, at 1903. The window is opened at the newly adjusted time at 1905 and closed at the adjusted time at 1907. The detect registers are updated at 1909 and a determination of a S/S word detect is made at 1911. If the S/S word is detected within the wider window, the digital SAT output is set high at 1913, the error threshold is set to 2 at 1915, and the program returns to the synchronization maintenance detection subroutine of Fig. 18. A lack of a S/S detect at 1911 causes the subroutine to return to the expanded window opening time step of !90S unless a three second timer has timed out at 1917. If the three second timer has timed out the program returns to ths SAT detect subroutine at

1605.

An interrupt driven background subroutine to handle S/S word detects is continuously running and is shown in Fig. 20. If the S/S detection window is open, as determined at 2001, a determination of whether a S/S or a S/S word detect has occurred at 2003 is made. A positive detect results in the window timer counter being reset around the last S/S ward detect time at 2005 and the S/S word detect is saved at 2009. If the window is not opened as determined by step 2001, a determination of whether S/S detection is made at times other than the S/S window time at 2011. S/S detection at this time means that digital signalling tone (DST) is being transmitted and 5 S/S detections result in a DST detect determined during the previously described program.

The subscriber unit maintains a second background subroutine which is.shown in Fig. 21. The subscriber may input commands to the subscriber unit via a switch, a keyboard, or other method which results in the subscriber unit taking a certain action. Additionally, he radio system may impose requirements on the transceiver which are communicated via the radio channel and conventionally interpreted by the radio transceiver logic system. The radio transceiver scans the inputs for possible commands at 2100 and determines whether the command is a user input or a requirement to terminate the call (DST transmit); or whether a handoff is being commanded by the system, other commands may be defined as necessary. (For clarity, the scan radio state, which may be caused to loop upon it self while waiting for an input, is shown in following flowchart with only single command output although any command output may be generated.

On. commanded DCSC transmit process is shown in Figure 22. Th. particular SAT frequency is specified by th. two bit cod. input to th. program by a hardcoding at th. fixed sit. or by th. received and detected cod. at the subscriber unit. This two bit code is input at 2201 and a S/S correlator word corresponding to this SAT frequency is selected from memory. Th. subscriber unit transceiver radio interface with t he subscriber is scanned for entry of additional commands at 2203. IF a request for digital privacy speech is made and input by the subscriber, the subscriber unit transmits a digital privacy speech request in conventional flash fashion on the reverse voice channel to the fixed site at block 2205. The subscriber unit waits in the scan radio state until a halt mode signal (shown in Figure 10) is received at 2207. Upon detection of the hold mode from the fixed site, the halt mode S/S sequence is recalled from memory (shown in Figure 10) at 2209 and caused to be transmitted at 2211. The subscriber unit remains in the transmit subroutine 2211 loop, transmitting one frame on each pass through subroutine 2211 until the digital privacy mode data signal is transmitted by the fixed site and detected at 2213. When the digital privacy speech mode transmitted by the fixed site is detected at 2213, the subscriber unit recalls the digital privacy mode from memory, at 2215, and again enters a transmit subroutine scan radio state loop. Each pass of the loop results in a transmission of a frame of digital privacy speech and a S/S word. When the subscriber elects to leave the digital privacy speech mode, his input is detected at scan radio 2217 and the request is transmitted to the fixed site via the halt mode pattern shown in Figure 10. A sequence of events similar to those shown in Figure 22 then occurs to remove the system from the digital privacy speech mode.

Th. transmit subroutine is shown in greater detail in Figure 33. The selected first S/S word of the system node sequence, determined at an earlier stage of the program, is input at 2301 and the first 504 data bits are input at 2303. The S/S word and the data bits are then transmitted sequentially at 2305 and 2307 respectively. Before returning to the main program, the selected system mode word sequence is rotated one word at 2309 so that the next S/S word is in the proper memory location for the next pass through the transmit subroutine.

If the scan radio state determines that a digital signalling tone (DST) must be transmitted such as when the radiotelephone call is to be terminated, the subroutine of Figure 24 is entered. The number of DST frames requires is obtained at 2401 and the current mode S/S sequence is recalled at 2403. The next S/S word sequence is recalled at 2403. The next S/S word in the sequence is transmitted at 2405 followed by 24 S/S words at 2407. The next S/S word in the sequence is rotated into place at 2409 before a test of the number of frames transmitted is made at 2411. If the last frame has been transmitted, the subroutine returns to the scan radio state.

Detection of a S/S word, in the preferred embodiment, is accomplished with a conventional correlator such as that shown in Figure 25. A known word (which may be a fixed 21 bit word in the fixed site or a 21 bit SAT/SYNC word identical to that decoded by the subscriber unit on the forward set up channel) is input to register 1 (2501). The received data is serially clocked into Register 2 (2502). The outputs of registers 1 and 2 are compared location by conventional exclusive OR functions whose outputs are checked for mismatch. If the number of mismatches is over a predetermined threshold, no detect is found.

Mod. detection say be accomplished by Inputting a binary logic level corresponding to a detected S/S word and the opposite binary level corresponding to a detected S/S word. The register 2601 in Figure 26 nay then 5 contain a sequence of bite corresponding to the S/S or S/S words detected. These bits, in ths preferred embodiment, form an address to a plurality of storage locations in ROM 2603. As the bits art clocked through register 2601, addresses such es those shown in the table 10 of Figure 26 are generated and are interpreted as the modes: digital data, digital halt, digital speech, or digital speech (RF only).

System information can be transmitted by verying the sequence of the aforementioned synchronization correlator 15 words. These sequences of sync (S) and inverse sync (S) words can also be though of as sequences of ones (1) and zeros (0). In the present invention they are treated as such. Like conventional linear block coding theory, these synchronization word sequences can be sectioned 20 into blocks with a minimum Hamming distance and can utilize conventional block codes. However, unlike conventional codes, the received sync sequence can have correct, incorrect, or missed sync words. Therefore, it is the task of the decoder of the present invention to 25 correct and/or detect both sync word sequence errors and misses to properly decode the correct sequence.

The coding structure of sync sequences can best be understood by using a vector diagram as shown in Fig. 27. Vector graph representation is useful, because it can 30 easily show the error and miss correction and detection capability of a sync sequence code. In order to keep the vector diagram only three dimensional, a simplified coding sequence is used. The vector points of the diagram represent the possible sync code sequences.

־ 32 Th.. sequences are repeated continuously to represent a particular system state. For example, the sequence S-S-S (point 3701 in Fig. 27) is transmitted as shown in Fig.

28. The figure showe that this sequence is made up of three cyclic sequence blocks. Like conventional block codes, the coding of this sequence block is kept cyclic with a minimum distance between sequence blocks. Unlike conventional block coding, the continuous (repetitive) transmission of a sequence block, combined with its cyclic property, maps all of the cyclic shifts of this sequence block into the same transmitted seuence or system state. For example, if the sequence is picked up at paint λ in Fig. 28, the receive sequence block is

S-S-S. This is a cyclic shift of the sequence S-S-S. Thus, the three cyclic sequence blocks that make up this particular system state are: S-S-S, S-S-S, and S-S-S.

By combining all cyclic shifts of a sequence block into one system state, the recovery of this state can be initiated at any point of the received sequence. This greatly decreases the time to recoever a decodable sequence block.

The sync sequence code represented by the indicated vector points in Fig. 27 are:

System State 0: Ή-S-S

S-S-S

S-S-S

System State 1: S-S-S

These sequence blocks which make up this sync sequence code have a minimum distance of 2. This can be seen on the vector graph in Fig. 27 as the distance between the indicated vector points of state 0 and state

1.

According to conventional coding, sync word errors map from one sequence block to another. For example if

S-S-S is transmitted and S-S-S is received; this single sync error can be represented as moving from vector point 2701 to vector point 2703 in Fig. 27. (With a minimum distance of 2, this sync error can be.detected but not corrected).

Missed sync words, however, require a new twist to vector graph representation: allowing the vector points to enter the planes and space between the sequence block points. This can be shown as follows on Fig. 27. Starting at sequence block point S-S-S (at 2701 ), consider receiving a single sync word miss, ^-S-S-M. This is represented as point 2705 on the graph. Notice that this point is equidistant from sequence blocks

S-S-S and S-S-S. A second sync word miss to the second sync word (S-M-M received) will map into the plane (point 2707). This vector point is equidistant from 4 sequence blocks starting with S. This can be similarily shown for any of the sequence blocks on the graph. Receiving all misses maps into the center of the cube. This vector representation for sync word errors and misses extends to higher multidimensional vector graphs for more applicable sequence codes.

Sync word error correction and detection capabilities, (with no sync word misses) follow convential code formulas. (See the following Theorems 1.1 and 1.2) For example, if the sync word sequence code has a minimum distance of 7, then 3 sync word errors can be corrected, or up to 6 sync word errors can be detected.

However, when sync word errors are combined with sync word misses, there is a trade-off in the error and sync correction capabilities. This is evident in the combined sync word error and sync word miss correction capability Theorems 1.3 and 1.4. For example, if the sync sequence has a minimum distance of seven, then a received sequence block with 4 misses and up to 1 sync word error can be corrected. Similarly,.a received sync word sequence with either 1 or 2 missed sync words and up to 2 sync word errors can be corrected. Table 1 give the sync error and sync miss correction trade-offs for this code:

TABLE 1

SYNC MISS AND ERROR CORRECTION CAPABILITIES SYNC SEQUENCE CODE D-MIN - 7

Miss Error Correct

ו4

Theorem 1.1

E = [ (D - 1) /2], where [ ] denotes a round-off co the largest integer no greater than (D-1)/2. (No sync misses)

Theorem 1.2

Ed ־ (D-1) (No sync misses)

Combined sync errors and misses:

Theorem 1.3׳

E . [ (D - Mr - 1) / 2]

Theorem 1.4

M - D - 2Er - 1 where:

D “ minimum distance of sync sequence code E - sync error correction capability Ed - sync error detection capability 10 M = sync miss correction capability Mr = number of received sync misses Er = number of received sync errors

These are guaranteed error and miss correction and detection capabilities. However, these are only 15 theoretical capabilities. The actual decoding performance depends upon the decoder design. The above theorems will be used later in the decoder proofs.

Theorems 1.3 and 1.4 are an extension of conventional coding capability theorems, to include the 20 case of combined sync word errors and sync word misses.

Theorems 1.3 and 1.4 can be graphically demonstrated for a very simplified sync word sequence code with sequence blocks of and S-S-S, shown in Fig. 29 . Theorem 1.4 says that this sequence code 25 (minimum distance = 3) can correct two misses and zero sync word errors or one sync error and no misses. Starting at vector point 3-S-S, 1800, any two misses will map into one of the three plane points shown as points 2901, 2903, and 2905. All three points are 30 correctable to ^-^-S־^. However, a received sequence block with 1 miss and 1 error could result in vector point 2907 (S-S-M) which is not correctable, as forecast by Theorem 1.3 «nd Theorem 1.4. Misses cannot map one sequence block into another, as was shown 35 in Fig. 27.

At least three synchronization sequence decoder types may be employed in the present invention and are described hereinafter. Two of the sequence decoders use received sync word sequence and miss detect information 5 to address the closest valid sequence via either a ROM table or a conventional combinational circuit. These two decoder methods generally are useable only when the number of valid sync word sequences is small. A third sequence decoder decodes the received sequence without requiring an exhaustive test of each sequence. Since the number of misses that can be corrected can be greater than the [( D-1) / 2] number of correctable errors, it is possible to decode a received sequence into the wrong system state. A technique is therefore necessary for determining whether the decoded sequence is correct and will be shown in the third sequence decoder type.

In the first synchronization sequence decoder, the received־sync sequence is detected by a single correlator with two thresholds. For example, if a 21 bit 20 synchronization word has a threshold of two or fewer allowed bit errors, then the inverse synchronization word is detected if there are 19 or more errors.

The received detect/miss history is saved in a register (detect register) with the following coding:

Sync detect 1 ־־, Sync miss = 0. The received detect sync/sync sequence history is saved in a second register (sequence register) with the coding: Sync detect = 1, Inverted Sync detect « 0, Sync miss = 0. Table 2 shows a sync sequence code with a minimum distance (D) equal to two and equivalent system states.

TABLE 2 «

Sequence Block System State

00001

01012

1010

10013

1100 0110 0011

11114

According to Theorems 1.3 and 1.4, this code can correct 1 miss and zero sync errors within a sequence block of 4 sync words. One sync error.with no misses or two or more misses can be detected but not corrected.

Four bit sequence and detect history registers require a 256 state ROM for decoding. One possible configuration is shown in Fig30 .־. Here the sync correlators, 711, 713, provide outputs to detect register 3001 and sequence register 3003: sync word detection or miss is coupled as a binary Logic Level to detect register 3001 and sync word detection or inverse sync word detection or miss is coupled to sequence register 3003. Registers 3001 and 3003 may be implemented as a dual 4 bit shift register (MC74HC14015 or equivalent). The decoder ROM 3005 may be a conventional read only memory (such as an MCM6551bC43 or equivalent) which provides an 8 bit output corresponding to the decoded system state or detection of an uncorreccable sequence block. The following table 3 shows part of the decoder ROM 3005 fill:

ROM ADDRESS (Seq ./Detect)

TABLE 3

ROM OUTPUT* (System State)

<td colspan="2"> 0000/1111</td><td> 01</td><td> • . . .</td>

<td> 5</td><td> 1111/1111</td><td> 04</td><td> • . . .</td>

<td></td><td> 1101/1101</td><td> 04</td><td> 1 Sync miss</td>

<td></td><td> 0001/0111</td><td> 03</td><td> 1 Sync miss</td>

<td></td><td> 0000/1110</td><td> 01</td><td> 1 Sync miss</td>

<td></td><td> 0100/1100</td><td> FF</td><td> 2 Sync misses</td>

<td> 10</td><td> 1101/1111</td><td> FF</td><td> Sync error</td>

*Detection of uncorrectable sync error or miss = FF

For long synchronization words, the probability of a sync miss is much greater than a sync error. For this reason, a system designer might decide to correct for 15 sync word misses only and detect sync word errors. This is also the case, of course, for sync word sequence codes with a minimum distance of 2.The circuit in Fig. 31 , illustrating a second synchronization sequence detector technique, will correct 20 all misses up to the capability of the code as defined by Theorem 1 .4 with E=0. It will also detect sync errors up to the capability of Theorem 2.1.

Theorem 2.1

Ed ־ D - Mr - 1 where:

D = minimum distance of sync sequence code Ed = sync error detection capability Mr ־. received sync misses

In Fig. 31 , the output of sequence register 3003 , 30 which is equivalent to a selected sync word sequence (Rsel) with values for the missed sync words selected as 0 in the preferred embodiment, is exclusive-ORed with the output of a decoded sequence generator 3100, which may be an up-counter (such as an MC74HC161 or equivalent), by an exclusive-OR (EX-OR) function 3101.

function 3101. The decoded sequence generator 3100 output is a parallel equivalent of a decoded sync word sequence (Sdec). The resulting output from the EX-OR 3101 is a parallel word in which the number of 1’s is equal to the Hamming distance between Rs el and Sdec. The EX-OR 3101 output is, in turn, ANDed with the output of detect register 3001 (which output indicates the relative position of the selected sync word detection values) in conventional AND function 3103. In the preferred embodiment, each selected sync word position is indicated by a zero. The resultant output of AND 3103 is the Hamming distance between Rsel and Sdec, with the selected sync word detection values deleted (or masked) from the Hamming distance, and if all zeroes are found between Rsel and Sdec when masked, a correct decode has occurred. If an incorrect decode has occurred (as shown by a non all zero output from AND 3103) , the sequence block generator 3100 is incremented to the next possible sequence and the process is repeated. In this type of decoder no errors in the detected sync words can be corrected.

The output of detect register 3001 is also further processed in a miss counter 3105, which is a combinational logic circuit which outputs a logic 1 if more than D-1 zeros are held in the detect register 3001. Such a logic circuit for D=2 is shown in Fig. 33. If more than D-1 misses are counted, the output from miss counter 3105 is a binary 1. If che number of misses are equal to or less than D-1, the output from miss counter 3105 is a binary 0. The miss counter 3105 output is conventionally OR-ed with the output of AND function 3103 in OR function 3107 to enable the phase state of the correct code co be provided. The function of the circuit of Fig. 31, in־־summary, is to iterate through all the possible sync word sequences, comparing each iterated sequence to the received sync word sequence until a correct system state decode can be achieved within the error correction capability of the code.

As an example, assume Che sync word ,sequence code of Table 2 is used. The transmitted sequence is

S-S-S-S, and the received sequence is

S-M-S-S. (Same system state with one sync miss) The sequence register 3003 will hold 1000 and the detect register 3001 will hold 1011. With a first iteration of a sequence of 1010 generated by the sequence block generator 3100, the EX-OR circuit 3101 output is 0010 and the AND circuit 3113 output is 0010. This is not the all zero state necessary for a correct decoding of the received sync word sequence; the sequence is then iterated. With a sequence of 1100 generated the EX-OR 3101 output is 0100 and the AND circuit 3103 output is 0000-a correct decode of system state 3.

The third synchronization sequence decoder of the preferred embodiment will correct all combinations of errors and misses within the bounds defined by Theorems

1.3 and 1.4 for all sync word sequence codes having a cyclic minimum distance D. The criteria for determining if the decoded system state is correct or incorrect is presented in the following three proofs. The Hamming distance between two sequences, SI and S2 will be denoted as d(S1, S2). (Distance is defined as the number of elements which differ between two linear codes). The Hamming distance is computed with the coding: S^s1, ^-5=0, and with sync misses selected as either 1 or 0.

Theorem 3.1:

Let Rsel be a received sync word sequence with all misses selected to maximize the Hamming distance between the transmitted (St) and selected (Rsel) received sync sequences, and also minimize the Hamming distance between Rsel and an incorrect sync sequence (system state). Given Mr received sync word misses, Theorem 1.3, E ^β [ (D - Mr 2 / (1 ־], defines the number of sync errors, E, which can be corrected for this code.

If the decoded sequence (Sdec) is an incorrect sequence, then d(Rsel, Sdec) > E.

St =* Transmit:ted sync word sequence

Sr - Received sync word sequence

Rsel = Selected sync word sequence (with misses in

Sr selected)

Sdec ^כ Decoded sync word sequence

Proof:

Utilizing the triangle inequality (see S. Lin et al.

Error Control Coding, Prentis-Hall, 1983, page

63) : d(v,w) + d(w,x) > d(v,x)

1. d(St, Rsel) + d(Rsel, Sdec) > d(St, Sdec)

Rearranging:

2. d(Rsel, Sdec) > d(St, Sdec) - d(St, Rsel) ־

For a worst case selection of missed sync detects,

3. d(St, Rsel) - Mr + Er - M + E

All misses are incorrectly selected.

Both Mr and Er have the maximum possible values consistent with a correct decode as defined by

Theorems 1.3 and 1.4.

i.e. Mr ³ M, Er = E

This also minimizes d(Rsel, Sdec) , for Sdec being an in correct sync sequence decode.

For an incorrect system state decode,

4. d(St, Sdec) = D

Substituting equation 3 and equation 4 into equation gives:

5. d(Rsel.Sdec) > D ־ (Μ + E)

Substituting Theorem 1.4 (M ־. D - 2E - 1) into equation 5 gives:

6. d(Rsel,Sdec) > D - (D-2E-1 + E)

The received sequence is bounded by che sync word error and miss criteria of Theorem 1,4.

Rearranging:

7. d(Rsel,Sdec) > E+1

8. d(Rsel,Sdec) > E ,

This is the lower bound, even with the selection in equation 3, which minimized d(Rsel,Sdec).

The above proof gives a lower bound for the Hamming distance between the received selected sequence, Rsel, and an incorrect decoded sequence Sdec. If a non over15 lapping upper bound for a correct decode can be determined, then the criteria for a sync sequence decoder is established.

Theorem 3.2

If the missed sync words of the received sync sequence and the corresponding sync words of a correctly decoded sequence are masked, then the distance between the two sequences is less than or equal to E. Where E is derived from Theorem 1.3.

d(Rsel.Sdec) masked £ E where

E ־ [ (D - Mr - 1) / 2 ] and Sdec is correctly decoded.

Proof:

For a correct decode of the received, sync word sequence with selected missed sync words, the distance

1. d(Rsel.Sdec) &lt; [ (D-1) / 2 ]

Rearranging Theorem 1.4 (D = M + 2Er + 1). Let Er have the maximum possible value consistent with a correct decode as defined by Theorem 1.3 ie. Er . E,

2. and D ־» M + 2E + 1.

Substituting equation 2 into equation 1 gives:

3. d(Rsel.Sdec) &lt; [ (M + 2E) / 2 ]

For a masked difference:

4. d(Rsel,Sdec)m &lt; [ 2E/2 ] = E d(v,w)m denotes that the distance is determined 15 after the missed sync words are masked, and have no bearing on the distance calculated.

5. d(Rsel,Sdec)m &lt; E

Therefore, by masking the received sequence and the decoded sequence prior to determining the distance, a 20 criteria for a correct sync sequence decode exists which does not overlap the criteria for determining an incorrect sync sequence decode. Masking, as the term is used herein, is the process of determining the value of the Hamming distance between the selected sync word 25 sequence (Rsel) and the decoded sync word sequence (Sdec) with the substituted values of the missed words removed from the calculation. For example, if the transmitted sync word sequence St = UUOOOOO, and the received sync word sequence Sr = MM11000, a selected sync word sequence 30 Rsel could be selected as 1111000. This Rsel would be decoded as the decoded sync word sequence Sdec of 1111111

- a totally .erroneous system state. Masking the selected values for the missed sync words yields an (Rsel)_m

11000. Comparing (Rsel)_m to a masked (Sdec)_Q of

11111 yields a Hamming distance of 3 ³ d(Rsel, Sdec)_m.

Calculating the value of E from theorem 1.3 and the fact that minimum distance between system states (D) in this example equals 7 and the number of received sync misses (Mr) equals 2, yields E-2. Thus, d(Rsel, Ddec)_m >

E»2 which violates the inequality of equation 5 of theorem 3.2 and proves that the decoded sync word sequence is incorrect. A new set of values for the missed sync words must be substituted and the Hamming distance recalculated for a new decoded sync word sequence. When the inequality of equation 5 is satisfied, a correct decode has been achieved.

Masking the misses before determining d(Rsel,Sdec) for the case in which an incorrect decode is made does not change the results of Theorem 3.1. This is because 20 the misses that are masked from equation 3 of Theorem 3.1 are also masked from the overall distance D in equation

4. See Theorem 3.3 and the following proof.

Theorem 3.3

IF the decoded sequence (Sdec) is an incorrect sequence, then d(Rsel,Sdec)m > E.

where E = [(D-M_r)/2]

Proof:

Utilizing the triangle inequality: d(v,w) + d(w,x) > d(v,x)

1. d(St,Rsel)m + d(St,Sdec)m > d(St,Sdec)m

Rearranging:

2. d(Rsel,Sdec)m > d(St,Sdec)m - d(St,Rsel)m

For a worse case selection of missed sync word detects,

3. d(St.Rsel) - Mr + Er = M + E

All misses are incorrectly selected. Both Mr and Er have the maximum possible values consistent with a correct decode as defined by Theorems 1.3 and 1.4. ie. Mr ^a M, Er . E This also minimizes d(Rsel,Sdec),

Rearranging Theorem 1.4; (D-M+2E+1) The distance with misses masked becomes:

4. D - 2E + 1

For an incorrect system state decode,

5. d(St,Sdec)m ־־ D(masked) 2 ־־E+1.

Substituting equation 3 and eq. 5 into eq. 2 gives:

6. d(Rsel,Sdec)m > 2E + 1 - E.

The received sequence is bounded by the sync word error and miss criteria of Theorem 1.4.

Rearranging:

7. D(Rsel,Sdec)m > E+1

8. d(Rsel,Sdec)m > E

Based upon the results of Theorems 3.2 and 3.3, a novel sync word sequence decoder process such as that shown in Fig. 32 can be implemented to correct for combinations of both sync word errors and misses. This decoder operates in conjunction with digital cellular signalling control (DCSC) functions 503 and 523 (shown in

Fig. 5) and״ more specifically may be implemented in the preferred embodiment as a subroutine to the instruction set of microprocessor 701 (Fig. 7). The sync word sequence decoder of Fig. 32 iterates through possible miss sequence possibilities, until the correct decode is confirmed but, significantly, does not need to iterate through all possible sequences thereby realizing a substantial savings of iteration time. The Decode N sync sequence block 3201 utilizes any conventional cyclic decoder. The received sync sequence loaded into this decoder is represented as a series oil's and 0’s, with the missed sync word value selected as a 1 or 0. In one embodiment che value is arbitrarily or randomly selected. In a second, the value is based upon an examination of the word and a best guess approximation is used to select the substitution values. Either technique may be employed in the present invention.

The possibility of the decoder mapping into the wrong sequence comes about because the number of correctable sync misses in Theorem 1.4 is greater than the sync error correction capability (Theorem 1.1) of the sequence code. For example, if D=7, and there are no sync errors, up to 6 sync word misses can be corrected.

(If greater than 6 misses, in this example, are detected, the decoder process is aborted at 3202). This is much larger than the Theorem 1.1 result of 3. When the decoder incorrectly decodes the receive sequence, the test of d(Rsel,Sdec)m > E at 3203, indicates that the decoded sequence is not correct and returns for a new set of detect values substituted for the missed syncs at 3205 and the missed sync values are saved with the sync detects at 3206.

For example, again assume a receive sync sequence 35 with a minimum distance of D . 7 between system states .

For this sequence code, the correctable combinations of sync mi..! and sync error. is shown in Table 1 (derived from the value of E of Theorem 1.3.) Assume a received sequence block arrives with 2 sync misses and either 0, 1 or 2 unknown sync errors. If the sync misses and either 0, 1 or 2 unknown sync errors. If ths sync misses are incorrectly seelcted to cause an incorrect decode into the wrong sequence stats, then ths resulting masked Hamming distance beween the received sequence and the decoded sequence, d(Rsel,Sdec)m determined at 3207, will be greater than I 2 ״ as determined from the £ value calculation of 3209. If this distance is equal to or less than 2, then the decoded sequence is the correct sequence. That is, all sync word errors and all incorrectly selected sync word misses have been corrected.

Thus, a signalling system for digital speech/data transmission via a cellular radio telephone system has been shown .and described. This system allows the. transmission of 9.6 kilobit digital privacy speech and high speed data through a cellular system while providing a digitally encoded signal for maintaining cellular system control such as SAT, signalling tone and control messages for handoffs, power changes, and other system functions. Efficient use of channel overhead capability and substantial reliability over a fading or noisy communications channel is realized from the use of 21 bit high autocorrelation, low self-correlation synchronization words and from the use of a novel error detection and correction technique which employs the fact that a synchronization word may be missed without being decoded as a wrong word. A substitution of a selected synchronization word detected value for a missed word detected value in a sequence of synchronization words detections enables a decoding function to test the sequence for Haiming distance between the sequence and possible correct sequencec. If the Hanning distance is less than or equal to the synchronization error correction capability of the coding function, a correct sequence is accepted. If the Haiming distance is greater than the synchronization error correction capability value, another selection of substitute synchronization word detect values for missed words is attempted until a decoded sequence with an acceptable Hamming distance is 10 realized.

Therefore, while a particular embodiment of the invention has been described and shown, it should be understood that the invention is not limited thereto since modifications may be made by those skilled in the 15 art. It is therefore contemplated to cover by the present application any and all such modifications that fall within the true spirit and scope of the basic underlying principles disclosed and claimed herein.