Digital Data Service System

Dataphone Digital Service (DDS) is a private line digital data point-to-point, or point-to-multipoint communication service, widely deployed in the United States of America and is described in detail in the following references:

1. "D4 Digital Channel Bank Family", Bell System Technical Journal, Vol. 61, Number 9, Part 3, November 1982.

2. "Digital Data System", Bell System Technical Journal, Vol. 54, Number 5, May-June 1975.

3. "Generic Requirements for the Subrate Multiplexer", Issue 1, TA-TSY-000189, Bell Communications Research Inc., April 1986.The data rates at which a DDS customer can obtain service are the subrates 2.4 kbps, 4.8 kbps, 9.6 kbps and the full rate, 56 kbps.

Service is deployed using a HUB architecture. As shown in FIG. 1, customer signals using Customer Service Units (CSU) or Data Service Units (DSU) deployed at customer premises 10 and served off end-offices 14 are picked up on 4-wire local loops 12. In the end-office 12, an Office Channel Unit (OCU) 16 interfaces the local loop 12 to a DSX-.0. cross-connect frame 18. Even though the loop signal is at one of the subrates, or the full rate, the OCU changes the rate to 64 kbps on the DSX-.0. side. The format of the 64 kbps is called the "DS.0.-A" format and is described in detail in Reference 2 above.

The signal, after passing through the DSX-.0. cross connect frame 18, is interfaced to a digital carrier facility 22 by a digital channel bank 20 through a plug-in unit, called a DS.0.-Dataport 24. In some applications the OCU 16 and the DS.0.-Dataports 24 are combined together into a plug-in unit called an OCU-Dataport. The digital carrier facility 22 connects the end-office 14 to a HUB office 26 where another DS.0.-Dataport 28 interfaces the DDS signal from the end office, which is buried in the carrier, to the DSX-.0. cross-connect frame 30 in the HUB office 26.

At this cross-connect, the signal is again in the standard DS.0.-A format. If the service is at subrates, the DS.0.-A signal is coupled to a Sub-Rate Data Multiplexer (SRDM) 32 where it is multiplexed with other DS.0.-A subrate signals to form a "DS.0.-B" signal, still at 64 kbps. This format, for various subrates, is also described in References 1, 2 and 3.

A full-rate (56 kbps) DS.0.-A signal cannot be further multiplexed at the DS.0. (64 kbps) level. The subrate DS.0.-B signals and the full-rate DS.0.-A signals are then coupled into second level multiplexers 34 (T1DM, T1BW4, D4 channel banks, etc.) and hauled onto the DDS network 36 of which the HUB 26 is a node.

DDS guarantees 99.5% Error Free Seconds (EFS) and 99.96% availability averaged long term. Inside the network, facilities used are carefully characterized for error performance and are protected with hot standby spares to ensure high availability. As a result, bandwidth inside the network is expensive, which is the reason why subrates are multiplexed into DS.0.-B signals before entering this network.

The digital carrier facility 22 between the end-office 14 and the HUB 26 is the weak link in the typical communication system thus far described. This facility is shared with other services (e.g., voice trunks) between the end-office and the HUB. These other services do not require the precision of DDS. Typically, these facilities are metro-exchange facilities and are designed for a Bit Error Rate (BER) of 10.sup.-6. The performance can drop to a BER of 10.sup.-3 before maintenance alarms are enabled.

To meet the 99.5% EFS criteria, the facility must run at a BER around 10.sup.-8. Hence, dataports in these facilities have to have Forward Error Correction (FEC) built into them so that after error correction, the effective BER is better than 10.sup.-8, even though the raw BER the facility is running at has dropped to 10.sup.-3. For subrates, the intrinsic redundancy in the DS.0.-A signal leads to an adequate error correction strategy based on majority voting. For full-rate DS.0.-A signal, a 64 kbps (DS.0.) parity channel is formed using a (16,8) shortened version of the well-known (17,9) Golay (BCH) code. The DS.0. parity signal is transmitted in another channel along with the full-rate DS.0.-A signal. The receiver at the far end of the facility 22 detects both the full rate DS.0.-A signal, as well as the parity-DS.0. signal, and corrects enough errors to ensure adequate performance. Details on FEC for both sub-rates and full-rates can be found in Reference 1.

Telephone companies are attempting to induce analog private line customers to change to DDS.0. service, which offers better quality data service (99.5% EFS and 99.96% availability). However, these customers use equipment operating at 2.4, 4.8, 9.6 and 19.2 kbps data rates. Standard DDS provides all these rates, except 19.2 kbps.

Accordingly, a need exists for simple and reliable method and apparatus for data transmission over existing communications facilities, as described above, at a data transmission rate of 19.2 kbps. The system should be capable of achieving a residual error rate after correction of 10.sup.-8 and the service should be transparent, as far as the existing central office circuits and error-free maintenance practices are concerned.

The invention comprises a method and apparatus for formatting digital information signals for transmission over existing communication systems at a frequency of 19.2 kbps. Three formats are utilized; one for local loop transmission over four wire loops, a second format for use between data ports incorporates an error correction capability, and a third format for use at the HUB office to allow DSX-.0. level entry into the standard DDS network hierarchy.

The local loop format employs Alternate Mark Inversion (AMI) encoding to convert digital data signals generated at the customer premises to an AMI line code format. In AMI, a ONE bit is transmitted as a pulse transition (polarity change) and a ZERO bit as no transition. For 19.2 kbps transmission, the period T of the bit is 1/19200 seconds. The pulse amplitude A is preferably 1.66 V.+-.6% to minimize excessive crosstalk with other digital services.

The second format, for communication between DS.0. dataports, comprises 40 bit words divided into 5 bytes, each byte of 8 bits length. The first byte in a word (DB1) comprises the sequence of a first framing bit F1, followed by 6 data bits (B1-B6) and a first control bit C1. The second byte (DB2) in a word comprises the sequence of a second framing bit F2, followed by 6 more data bits (B7-B12) and a second control bit C2. The third byte (PB1) in a word comprises a sequence of eight parity bits used for error correction of the DB1 byte. The fourth byte (PB2) contains eight more parity bits of the DB1 byte for error correction of the DB2 byte. The fifth byte is a word framing byte PB2 comprising the inverse of PB2.

The third format 19.2 kbps signal, for use at the HUB office for communication between a DSQ dataport and a DDS network, also comprises a 40 bit word divided into 5 bytes, each byte of 8 bits length. Byte one (UMC1) comprises the bit sequence of OXXXXXXX ("X" is a don't care bit). Byte UMC1 is taken from a first unassigned MUX channel; bytes two and three comprise DB1 and DB2, respectively, wherein the first bits of each are forced to 1 if in error. Bytes four and five comprise 8 bits each from UMC's 2 and 3. The code sequence for each is OXXXXXXX, as in UMC1. The first bit in each byte in the sequence of bytes 1-5 follow the pattern 01100 which is the correct format for a multiplexed 9.6 kbps DS.0.-B signal (See References 1-3).

The three formats described above enable a 19.2 kbps signal to be transmitted over local loops, as well as unqualified loops with level entry into standard DDS network hierarchy, while still enabling error correction capability, transparency and normal maintenance.

The invention will now be described in detail in connection with the drawings.

FIG. 1 is a block diagram of the typical architecture of an existing prior art DDS communication network.

FIG. 2 is a block diagram of transmitter/receiver system employing AMI modulation signalling over four wire loop between, for example, a customer's premises and end-office.

FIG. 3 is a timing diagram showing the time slots allotted in line (a) for local loop signalling at 19.2 kbps, line (b) for DS.0. format and line (c) for DS.0.-A format.

FIG. 4 is a block diagram of the error correction encoding system of the invention for 19.2 kbps signalling.

FIG. 5 is a block diagram of the error correction decoding system of the invention for 19.2 kbps signalling.

As shown in FIG. 2, customer digital data is encoded at the customer premises 10 in the form of Alternate Mark Inversion (AMI) modulation, using well established techniques (See, e.g., U. S. Pat. No. 4,220,816). The AMI data is transmitted by transmitter 40 at a rate of 19.2 kbps to a receiver 44 at an end-office 14 through a shaping filter, such as a 3rd order Butterworth Filter (See Signals & Systems by A. V. Oppenheim et al., pp 424-425 Prentice-Hall, Inc. 1983). The real pole 42 of the Butterworth Filter is located at transmitter 40 and is coupled across one pair 46 of the four wire loop, while the complex pole 48 is coupled across the pair 46 at the end-office 14 at the input to receiver 44.

As shown in FIG. 2, the amplitude A of the AMI bits transmitted is set at 1.66 volts.+-.6% and the bit period T is 1/19200 seconds. The waveform of the received signal s(t) at the output of the complex pole 48 of the Butterworth Filter is shown on line 47.

It can be shown that at a transmission rate of 19.2 kbps, with A=1.66V.+-.6% and T=1/19200 seconds, the resulting transmission over 2 wire lines will have spectral characteristics that will not cause excessive cross-talk interference with other digital services. Although standard DDS engineering rules require local loops to not exceed 34 db loss at Nyquist frequency (9.6 KHz for 19.2 kbps data rate), it can be shown that with the format described above, adequate margins can be maintained on 1% Near End Cross Talk (NEXT), even when the loop has up to 43 db loss at Nyquist Frequency.

The incoming AMI signal at 1.92 kbps at RCVR 44 must now be framed for transmission from the DS.0. dataport 24 of end-office 14 to the DSQ dataport 28 at HUB office 26 via Digital Carrier Facility 22 (See FIG. 1). As noted earlier, the OCU 16 converts the rate of all incoming signals to 64 kbps in what is called a DS.0.-A format prior to interfacing with the digital carrier facility 22. The error correction format of the present invention must be made compatible with the DS.0.-A format. This is achieved, as shown in FIG. 3, by forming 40 bit words --i-1, i, i+1--, etc. at the receiver 44 (FIG. 2) or DS.0. dataport 24 (FIG. 1). Each 40 bit word is divided into 5 bytes of 8 bits length labelled DB1, DB2, PB1, PB2 and PB2.

Byte DB1 consists of 6 data bits taken from the 19.2 kbps local loop signal; to which a 1 bit is prefixed and a control bit C is appended. Similarly, byte DB2 consists of the next 6 data bits in the 19.2 kbps local loop signal, to which a 1 bit prefix is added and a control bit C is appended. The control bit functions are described in Reference 2 for prior DDS rates and remain unchanged for 19.2 kbps. Note that a secondary data channel can be safely buried in the "C" channel without creating an undesirable 0000000 byte, because by definition, each byte has a prefix of 1. Next, 8 bit parity bytes PB1 and PB2 are formed, as described in Reference 1 pp. 2750 through pp. 2756, using the (16,8) shortened Golay Code for error correction purposes. PB1 is the parity on DB1 and PB2 is the parity on DB2. A fifth 8 bit byte equal to PB2 (every PB2 bit inverted) is formed. In this manner, successive 40-bit words are formed, consisting of the five bytes DB1, DB2, PB1, PB2 and PB2. These words are transmitted from end-office 14 via digital carrier facility 22 at the DS.0. (64 kbps) rate. Comparing line a) of FIG. 3 with line b), it can be seen that 40 bits at 64 kbps, line b), takes 625 us, just as the 12 bits at 19.2 kbps, line a). It can be seen, therefore, that this is a compatible format.

The HUB office receiver at DS.0. dataport 28 frames the 40-bit word transmitted over the digital carrier facility by demodulating the signal and comparing incoming 8 bit bytes and detecting inverted adjacent bytes. This comparison must be "flywheeled", i.e., an N out of M type test (See Reference 1) must be done before declaring framing to ensure high framing reliability, even in the presence of BER around 10.sup.-8. It can be demonstrated for reasonable numbers like N=16, M=20 word framing error events will be separated by several years.

Having framed the word, the receiver corrects errors on DB1; using PB1; and on DB2; using PB2, as described in Reference 1, pp. 2750 through pp. 2756. It then converts the error corrected DB1 and DB2 into the DS.0.-A format, as described in the next section. This DS.0.-A signal is then coupled to the DSX-.0. cross-connect frame 30 for routing, as discussed earlier.

FIG. 3, line C, shows the DS.0. format for 19.2 kbps at the HUB office 26. Again, a 40 bit word is formed using 5 bytes. The first byte has bit 1, the first bit transmitted, set to a "0" the error correcting logic and maintenance circuits of the system are insensitive to the state of the next 7 bits in the byte, which are labelled "X" (don't cares). This takes up a time interval of 125 usec. The DB1 and DB2 bytes occupy the next two 125 usec. time intervals. The first bit of DB1 and DB2 are error corrected by error correction, if needed. Two more bytes, similar to the byte in the first 125 .mu.sec. time interval, are appended to form the 40 bit word. The rate is readily shown to be 64 kbps. The entire 5 byte word takes up a 625 .mu.sec. time interval. It can be seen in FIG. 3 line C, that the pattern formed by bit 1 of each byte is 01100, which is the correct format for a multiplexed 9.6 kbps DS.0.-B signal (References 2,3). This 19.2 kbps DS.0. signal of FIG. 3, line C, can be viewed as a partially filled DS.0.-B subrate signal,

In order to be compatible with other current DDS maintenance procedures, it is advantageous to repeat bit 2 through 8 of DB2 in the "X" (don't care bits 2 through 8) of the unused bytes 1, 4 and 5. This is because some maintenance systems send specific multiples of bytes which they expect will be seen by the unit which is to be tested. A detailed definition of these maintenance procedures is provided in Reference 4 ("Digital Channel Banks - Requirements for Dataport Channel Unit Functions", Issue 3, TA-TSY 000077, Bell Communications Research Inc., April 1986).

It should be noted that the 01100 frame pattern in bit 1 of the 5 bytes that make up the 19.2 kb/s word must be maintained. Also, since in the 19.2 kbps DSO format, the first bit of DB1 and DB2 is a "1", the control channel "C", discussed earlier, can be bit robbed to create a secondary channel. If, in every third data byte (DB1, DB2), the "C" bit is robbed, the rate of this secondary channel will be 16/15=1.06 kbps, where 6 means 6 recurring. In summary, use of the above-described signal formats enable a 19.2 kbps data rate signal to be added to DDS. These formats enable transmission at 19.2 kbps over local loops, as well as error correction over unqualified (with respect to BER) digital facilities. In addition, the 19.2 kbps DS.0. format, as defined, allows these signal DSX-.0. level entry into the standard DDS network hierarchy. The signals discussed can be readily maintained using existing DDS maintenance tools. Finally, secondary channel capability of up to 16/15 kbps has also been described.

FIG. 4 shows an implementation of 19.2 kbps encoding to allow later error correction. Serial Data of 64 kbps enters a serial to parallel converter 401 and is converted to 8-bit parallel data (indicated throughout by an 8/ at the line). An elastic store 402 removes any timing phase differences between the input data clock and the output data clock (not shown). The two data bytes are identified by a one in bit 1, and are stored in data registers (D.R.) 403 and 404, respectively. Multiplexer switch 405 selects data byte 1 from register 403 under control of timing logic control 406. Parallel to serial converter 407 then converts data byte 1 to serial and sends it out. Then, multiplexer 405 selects data register #2 (404), which is sent out. Next, multiplexer 405 selects data register 403. This data passes through BCH parity matrix logic circuit 409 (See Reference 5, "Error Correcting Codes", W. W. Peterson and E. J. Weldon, Jr., M.I.T. Press, 1972). Multiplexer switch 408 selects the parity matrix 409 output, sending data byte 1 parity to the parallel to serial converter 407. Multiplexer 405 then switches to data register #2 (404), which goes through the parity matrix 409 and the parallel to serial converter 407. Multiplexer 408 then selects the inverted parity from inverters 410 to send out the inverted parity of data byte 2.

FIG. 5 shows the 19.2 kbps error correcting decoding. Serial to parallel converter 501 accepts the 5 bytes frame DB1, DB2, PB1, PB2 and PB2. Framer 502 detects the fact that PB2 is the inverse of PB2 and in this manner identifies the frame boundaries. Data bytes DB1 and DB2 are stored in registers 503 and 504, respectively.

Parity bytes PB1 and PB2 are stored in registers 505 and 506, respectively. First, multiplexer switch 507, under the control of control logic 514, selects data register #1 (503), while multiplexer switch 508 selects parity register #1 (505). The data is encoded using the identical parity matrix 509, as was used in the encoder 409 of FIG. 4. The result is modulo-2 added (XOR-ED) to the received parity PB1 from multiplexer 508 by XOR-GATES 510. This output is referred to as the syndrome (See U. S. Pat. No. 4,382,300 to D. Gupta, issued May 3, 1983). If no errors occurred, the syndrome will be zero. If a correctable number of errors occurred, the syndrome will indicate which specific bit or bits should be inverted to correct them. This conversion is done by Error Correction logic 511. Next, multiplexer switches 507 and 508 select DB2 and PB2, respectively. In an identical manner as DB1 and PB1, errors are corrected. Elastic store 512 removes any jitter between input and output clocks. Parallel to serial converter 513 converts the 8-bit parallel data to a serial output bit stream.

While the invention has been described in connection with specific methods and apparatus, it is to be understood that the description is by way of example and not as a limitation to the scope of the invention as set forth in the following claims. For example, while specific reference has been made to the Dataphone Digital System of AT&T, in the introduction, the invention has applicability to digital data services, in general, wherein reliable 19.2 kbps signal transmission is required.