1、INTERNATIONAL TELECOMMUNICATION UNION)45G134 8 TELECOMMUNICATIONSTANDARDIZATION SECTOROF ITU05“,)#G0G0$!4!G0G0.%47/2+3.%47/2+G0G0!30%#43%.%2!,G0G002).#)0,%3G0G0(b) that some error correction procedures may be more advantageous than others depending on transmissiondelays in the network and on the dis
2、tribution (with time) of errors;(c) that the distribution (with time) of errors at the ends of a path may depend on forward error correctionprocedures implemented in one or more of the path segments;(d) that the applicability of some error correction procedures may be affected by the number of satel
3、litesystems in the connection, which may be in national or international links or in the Maritime Mobile Service;(e) that different error correction procedures may be concatenated in some connections,unanimously recommendsthat the general principles identified in this Recommendation be taken into ac
4、count in the design and applicationof procedures for the detection and correction of errors in public data networks.1 General1.1 The purpose of this Recommendation is to describe general principles applicable to the detection and correctionor recovery of link transmission errors in public data netwo
5、rks.1.2 Two fundamental objectives of error control procedures are:- to ensure an incidence of undetected errors that is within acceptably low probability limits;- to ensure that detected errors are corrected or recovered using an error control procedure consistent withdata throughput and sequencing
6、 requirements which apply when the error rate of the Physical Layer iswithin the fully acceptable and the tolerance limits of specified performance.1.3 In the context of the Reference Model of Open Systems Interconnection, it is noted in Recommendation X.200that each (N) peer protocol should include
7、 sufficient control information to enable the (N) entities to detect or recoverfrom error conditions within its purview. Reporting detected but unrecovered errors is a service that must be provided byeach layer.Specifically, it is an objective of the Data Link Layer to detect and possibly correct er
8、rors which may occur inthe Physical Layer.1.4 For any particular error detection arrangement the probability of undetected errors will generally tend toincrease:- with increasing error rate,- for any given error rate, as the error distribution becomes less random and as the length of error burstsinc
9、reases,- with increasing frame length,- possibly due to scrambling arrangements which may have factors in common with the generatingpolynomial used for error detection.2 Fascicle VIII.3 - Rec. X.1411.5 Data throughput in the presence of errors depends on the design of the error control procedure, wh
10、ich in turndepends on the following conditions:- error rate,- error distribution,- scrambling and/or multiplexing arrangements insofar as they affect the error distribution or error rate,- transmission path (propagation) time delay,- data signalling rate,- frame length,- window size,- buffer memory
11、resources at the sending and receiving end of the link.2 Types of error occurrencesError occurrences are typically of three types distinguished by characteristic error distributions with time:- random errors,- burst errors,- errors due to uncontrolled slip.It is likely that one type of error occurre
12、nce will be predominant in any particular link, depending on the type oftransmission systems employed (i.e. cable, microwave radio relay, or satellite, with or without forward error correction).In the design of error control procedures for a link, it is important to identify any tendency for the pre
13、dominanceof a particular type of error occurrence.3 Error control procedures3.1 Types of procedureTwo types of error control commonly employed in public data networks (PDNs) are:- forward error correction, a coding method employed with the objective of detecting and correcting errorsin received data
14、 instead of requesting retransmission,- ARQ procedures wherein transmitted information is formatted in frames with error detection encoding,and error recovery is achieved by automatic repetition upon request from the data receiver of a frame or ofall information already transmitted starting with the
15、 requested frame. Timeout recovery serves as a backupfor the ARQ procedure.3.2 Forward error correctionForward error correction (FEC) does not require the provision of a backward mechanism in order to operate.FEC is usually applied at the Physical Layer of the reference model, typically within trans
16、mission systems whose errorperformance might not otherwise meet required limits.The capability of FEC techniques commonly used in PDNs to control errors tends to be restricted to thecorrection of a limited number of errors (typically 2 or 3 errors) within each coded information block or constraint o
17、fblock length. For this reason these FEC procedures are most effective in situations where error occurrences arepredominantly random.Depending on multiplexing arrangements and sometimes on other arrangements in the Physical Layer such asscrambling and encryption, residual uncorrected errors after FE
18、C may tend to be grouped in clusters or error bursts.When the number of errors within a coded information frame or constraint length of code exceeds the correctioncapability of the FEC algorithm, the total number of errors in the cluster or burst may be increased rather than reduced bythe FEC facili
19、ty.Usually, it will not be feasible to notify the Data Link Layer of detected but uncorrected errors via FEC facilitiesof the Physical Layer which may perform their error control function at a multichannel, multiplexed signal level of thetransmission system.Fascicle VIII.3 - Rec. X.141 3In the adapt
20、ation of data signals at recommended bit rates below 64 kbit/s for transmission at 64 kbit/s,sufficient redundancy will be introduced in some cases for forward error correction to be undertaken on a majorityvoting basis without special forward error correction encoding. With this arrangement, a larg
21、e number of different errorpatterns can be detected and corrected.Alternatively, or in addition, the frame checking sequence of the ARQ error control procedure may also be usedto distinguish between correctly and incorrectly received information in redundant signal streams.3.3 ARQ procedures3.3.1 Ge
22、neralARQ procedures require the provision of forward and backward channels, usually with simultaneoustransmission capability.ARQ control procedures of error detection and error recovery are included in the functions of the Data LinkLayer and may also be implemented in the functions of higher layers
23、of the reference model.3.3.2 Error detection3.3.2.1 Frame checking sequenceThe 16-bit frame checking sequence (FCS) described below is used for error detection in the packet transferprocedures of Recommendations X.25 and X.75, in the Signalling System No. 7 signalling link procedure ofRecommendation
24、 Q.703 and in the Link Access Procedure on the D-channel of an ISDN as described inRecommendation Q.921 (I.441).The same generator polynomial is also used in the encoding and checking process of Recommendation V.41.The 16 FCS bits are generated at the transmitter. They are the 1s complement of the s
25、um (modulo 2) of:1) the remainder of xh(x15+ x14+ x13+ . . . + x2+ x + 1) divided (modulo 2) by the generator polynomialx16+ x12+ x5+ 1, where h is the number of bits in the frame existing between, but not including, the finalbit of the opening flag and the first bit of the FCS, excluding bits inser
26、ted for transparency, and2) the remainder after multiplication by x16then division (modulo 2) by the generator polynomialx16+ x12+ x5+ 1 of the content of the frame existing between, but not including the final bit of theopening flag and the first bit of the FCS, excluding bits inserted for transpar
27、ency.As a typical implementation, at the transmitter, the initial remainder of the division is preset to all 1s and is thenmodified by division by the generator polynomial (as described above) on the address, control and information fields;the 1s complement of the resulting remainder is transmitted
28、as the 16-bit FCS sequence.At the receiver the correspondence between the check bits and the remaining part of the frame is checked. If acomplete correspondence is not found the appropriate error recovery procedure is initiated.As a typical implementation at the receivers, the initial remainder is p
29、reset to all 1, and the serial incomingprotected bits including the check bits (after the bits inserted for transparency are removed) when multiplied by x16andthen divided by the generator polynomial will result in a remainder of 0001110100001111 (x15 through x0respectively)in the absence of transmi
30、ssion errors.Explanatory notes concerning the FCS error detection procedure described above are given in Appendix IThe procedure will detect:a) all odd numbers of errors within a frame,b) any error burst not exceeding 16 bits in length,c) all two-bit errors when the code length is less than 32768 bi
31、ts,d) a large percentage of other error patterns (with even numbers of errors).3.3.2.2 Use of scramblersThe following system design consideration should be taken into account concerning the use of self-synchronizing scramblers:4 Fascicle VIII.3 - Rec. X.141Where self-synchronizing scramblers (i.e. s
32、cramblers which effectively divide the message polynomial by thescrambler polynomial at the transmitter and multiply the received polynomial by the scrambler polynomial at thereceiver) are used, the scrambler polynomial and the generating polynomial for error detection must have no commonfactors in
33、order to ensure satisfactory performance of the error-detecting system. Where this condition cannot bemaintained, the scrambling process must precede the error detection encoding process and the descrambler process mustfollow the error detection decoding process. Where additive (i.e. non-self-synchr
34、onizing) scramblers are used or wherethe scrambling takes place at a multi-channel multiplexed signal level, this design precaution need not be observed.3.3.2.3 Frame integrityThe integrity of the frame format must be maintained in order to assure proper functioning of the error detectionprocedure d
35、escribed in 3.3.2.1.The frame structure for all transmissions is distinguished by opening and closing flags, each consisting of one 0followed by six contiguous 1s and one 0. A single flag may be used as both the closing flag for one frame and theopening flag for the next frame. To ensure that the un
36、ique flag sequence is not simulated, the entire frame contentbetween two flag sequences is examined at the transmitter and a 0 bit is inserted after all sequences of 5 contiguous 1 bits(including the last 5 bits of the FCS). At the receiver, the frame content is reexamined and any 0 bit which direct
37、lyfollows 5 contiguous 1 bits is discarded.At the receiver, a frame validity check is carried out to detect any invalid frames not properly bounded by twoflags or having fewer than the specified minimum number of bits. Invalid frames are treated in the same way as frameswith detected errors.3.3.3 Er
38、ror recovery proceduresIn accordance with ARQ concepts, error recovery is vested in the traffic control procedure wherein allinformation frames are numbered sequentially in order of transmission, from 0 through modulus minus 1 (wheremodulus is the modulus of the sequence numbers). Typically, the mod
39、ulus equals 8 or 128 and the sequence numberscycle through the entire range.Valid frames without errors received in proper sequence are acknowledged in responses from the receiver to thetransmitter, while invalid frames and frames with errors are discarded by the receiver and completely ignored. Fra
40、merecovery action is initiated by the receiver when a valid frame without errors does not have the expected sequencenumber. Consequently, when one or more frames are discarded for lack of validity or for errors, the number of the nextcorrectly received frame will be out of sequence, causing the rece
41、iver to initiate the prescribed frame recovery procedure.If, due to a transmission error, the receiver does not receive (or receives and discards) a single informationframe or the last in a sequence of information frames, then the out-of-sequence condition which would otherwise serve toinitiate erro
42、r recovery procedures at the receiver will not be detected. In this case, frame recovery will be initiated at thetransmitter via a time-out procedure as follows:For traffic control purposes, the receiver must send acknowledgment response to the transmitter confirming thereceipt of valid, error free
43、frames. After a specified time-out period with outstanding transmitted frames and noacknowledgement or frame recovery responses from the receiver, appropriate recovery action is initiated at thetransmitter to determine the point at which retransmission must begin.Alternative types of error recovery
44、procedure are available as follows:- reject procedure,- selective reject procedure,- selective reject-reject procedure.Each of these procedures requires that storage be provided at the transmitter for all information frames alreadysent but not yet acknowledged to be correctly received.The data throu
45、ghput efficiency obtainable as a function of error rate and distribution may depend significantlyon the type of error recovery procedure, particularly in transmission links with associated long time delays (e.g. links viasatellite). The complexity of error recovery implementation, including frame st
46、orage requirements at the receiver, isanother consideration which plays an important part in selecting the most advantageous error recovery procedure to suita particular situation.3.3.3.1 Reject (REJ) procedureThe reject (REJ) error recovery procedure is used by the receiver to request retransmissio
47、n of informationframes commencing with a specified sequence number and to simultaneously acknowledge satisfactory reception of allpreceding information frames.Fascicle VIII.3 - Rec. X.141 5The rejected frame and all subsequent information frames already in transit at the time that the REJ responsere
48、aches the transmitter will be retransmitted.After sending the REJ response, the receiver discards all incoming information frames until the lost frame isrecovered. This procedure minimizes frame storage requirements at the receiver, but under marginal conditions of errorperformance it may result in
49、poor throughput efficiency depending on the round-trip transmission delay between thetransmitter and receiver.With the REJ error recovery procedure, the window size should allow a maximum number k of outstandingframes, where k is the smallest integer not less than r, calculated as follows:whereT is the transmission rate (bit/s)D is the round-trip delay (seconds)L is the information frame length (bits).3.3.3.2 Selective reject procedureThe selective reject (SREJ) response is used by the receiver to request retransmission of a single informationframe identified by its sequence number and to
