CN101193060A - Method for reliable E1 transmission based on forward error correction mechanism in packet network - Google Patents

Abstract

A packet network reliable end-to-end E1 transmission realization method which adopts the FEC mechanism is provided. At the end where E1 enters the packet network, E1 data flow is stored in the buffer area (101) and then encapsulated into data packets (102) and then generates FEC packets. The encapsulated data packets and FEC packets are transmitted (103) on the packet network (107). At the end where E1 leaves the packet network, the receiving timer is set to control time delay and the de-dithering buffer area is set to eliminate the influence of the packet network time delay dithering. E1 data are extracted from the received data packets and are then stored in the de-dithering buffer area (104). FEC packets are stored in the error correcting buffer area. When the receiving timer overflows, the packets in the buffer area are checked, lost data packets are calculated and lost information (105) are restored. At the same time E1 clock is restored and E1 code stream (106) is restored according to the clock and the de-dithering buffer area. The invention can also guarantee remote seamless interconnection and intercommunication without FEC mechanism.

Description

On Packet Based Network, adopt forward error correction mechanism to realize the method that reliable E1 transmits

Technical field

The present invention relates to the communication means of Packet Based Network transmission E1, particularly relate to the FEC that on Packet Based Network, realizes transmitting the E1 business (Forward Error Correction, forward error correction, down with) method.

Background technology

(1.TDM Time-Division Multiplexing, time division multiplexing) technology

TDM is the abbreviation of time-division multiplex technology, is the technology that a kind of timesharing transmits a plurality of data flow, and it is time slot of each distribution of flows, and repeats to transmit a fixing sequence of time slots on same transmission channel.

The conventional telecommunications net adopts PDH and SONET/SDH system to provide the transmission service in the TDM mode, mainly provide voice service to the user, though this mode can provide highly reliable service, its cost height, cost an arm and a leg, shortcoming that limited bandwidth, the cycle of laying are long is also very outstanding.

On the other hand, the professional and application and development trend from current network, the SONET/SDH technology has had extensive application on the net through the development in twenties years, and the application scale has reached tens dollars.

E1 is the primary group of TDM.

2.IP and Ethernet

Based on IP with based on Ethernet is two kinds of packet transport technologies that are most widely used, but IP and Ethernet are in the different levels of network.IP is the network layer technology, and its immense success at Internet makes it be in critical dominant position in packet network, it should be noted that how IP carries on Ethernet.Ethernet is in below the network layer, because its ubiquity (the incoming data stream amount of carrying more than 90%) is simple and easy to characteristics such as usefulness and high performance-price ratio, essence is most widely used packet network technologies.

Existing network is trend of the times to Packet Based Network evolution, but network and the packet-based network based on E1 is simultaneous in the process of evolution.

Packet Based Network is voice-bearer simultaneously, and business such as video and data are necessary also being inevitable, and transmit with E1 and current speech business is many.Thereby be necessary to transmit E1 such as the Packet Based Network of IP and Ethernet.

3. the QOS such as the Packet Based Network of IP guarantees technology

QoS problem such as the Packet Based Network of IP is to solve Packet Based Network to carry multiple services key factor, from current achievement in research, mainly contains following several solution.

＜1〉IntServ (Integrated Service, integrated service)

At the problem of IP QoS, IETF has proposed intserv model in early days.Intserv model is called integrated service again, and its basic thought is before transmitting data, carries out network resource reservation based on moving speed according to the QoS demand of business, thereby QoS guarantees for this data flow provides end to end.For this reason, stream-oriented resource reservation protocol (rsvp) is adopted in integrated service usually, and each node on the flow transmission path is reserved and maintenance resources for stream.Host computer using RSVP proposes the request of QoS for application stream to network; Router utilizes RSVP the QoS solicited message to be passed to other routers in the path of stream, and sets up and preserve the information of this service; The RSVP request will make to be reserved at the router place along the resource of data path.The advantage of this model is that the QoS that can provide absolute end to end guarantees.

But this model is very difficult in realization, is mainly reflected in:

(1) be based on owing to reserving that each stream carries out, therefore make the state information that will keep each stream in the node, cause the core router burden too heavy, so extensibility to be very poor.

(2) each node all will be safeguarded types of databases in the network, and realizes complicated functional module (as resource reservation, route, admittance control etc.), has caused great complexity.

IntServ is applicable to network on a small scale.

＜2〉DiffServ (Differentiated Services, Differentiated Services, or segmentation business)

Because the limitation of IntServ, IETF has proposed the DiffServ model again, is called differentiated service again.The basic thought of differentiated service is to be each packet classification in the porch of network, and in packet the corresponding differentiated Services code point (DSCP of mark, DiffServ CodePoint), be used for designation data and wrap in mode processed on the intermediate node in forwarded path.In the core router of network internal, only preserve the response mechanism of simple DSCP and PHB (whenever slip a line into), according to the DSCP value in the packet head packet being carried out corresponding priority level transmits, and the realization of the preservation of Business Stream state information and flow control mechanism etc. are all carried out at the network boundary node, and internal node is irrelevant with state.

It is simple that Differentiated Services has realization, the characteristics of favorable expandability.At present Differentiated Services has obtained the support of most producers in IP network, and its specific implementation technology comprises classification, heavy label, rate limit, traffic shaping, Congestion Avoidance, queue scheduling etc.

But the limitation of Differentiated Services is also very outstanding, is mainly reflected in:

(1) Differentiated Services is only promised to undertake relative service quality, thereby can not provide absolute service quality to guarantee to the user.

(2) when congested generation, differentiated service can only be taked the mode of dropping packets, and can not adopt the mode of for example bypass that partial discharge is reached home by other paths.

(3) for the business of equal priority, equipment when congested to message to abandon right and wrong intelligentized, that is to say that equipment is dropping packets randomly, consequently all professional service quality all are affected.And the result that wish this moment is the message that only abandons the small part Business Stream, thereby avoids the service quality of the most Business Stream that is left to be affected.

＜3〉IntServ combines with DiffServ

At present industry has also proposed mode that IntServ is combined with DiffServ, its thinking is: still use RSVP at user network, the type of service of IntServ is mapped as the type of service of DiffServ at the DiffServ of operator network boundary, utilize the framework of IntServ to solve QoS (Qualityof Service end to end like this, service quality), the autgmentability of also utilizing DiffServ to provide simultaneously.

But this method still exist IntServ the signaling complexity, be difficult to problems such as management, and owing to network using DiffServ in operator, therefore also can only provide relative QoS at this section network, service quality can not get rigid bandwidth assurance thereby make end to end.This method still is in a kind of Study on Theory stage at present.

<4>MPLS(Multiprotocol Label Switching)& QoS

Utilize multiprotocol label switching MPLS technology, can help solve the QoS problem.MPLS is a kind of switching technology in conjunction with the second layer and the 3rd layer, has introduced the mechanism based on label, Route Selection and data forwarding separately, stipulates the path of a grouping by network by label.The MPLS network is made up of the LSR (LSR) of core, the tag edge router (LER) of marginal portion.

Because MPLS adopts label to exchange to carry out MPLS and transmit, so its forward efficiency is higher than the forwarding of Traditional IP by router, thereby improves QoS by reducing forwarding time.In addition, comprise the EXP field of a 3bit in the heading of MPLS, can this MPLS priority of messages of mark by this field, thus equipment can be treated with a certain discrimination according to Priority flag when transmitting this MPLS message.

The limitation of this mode is: at first it must be realized based on the MPLS network, and does not implement MPLS on current many networks; Along with the continuous development of chip technology in recent years, the performance difference between routing forwarding and exchange are transmitted is also more and more littler in addition; And carry out the implementation that prioritization in fact also is DiffServ, thereby this mode some limitation of also having DiffServ inevitably and being had by EXP.

<5>MPLS-TE & QoS

Traffic engineering (TE:Traffic Engineer) is meant the processing procedure of selecting the path into Business Stream, with balance Business Stream load between links different in network, router and the switch.Its target is to give calculating one paths (source routing) between certain node and another node, and its constraint (for example bandwidth/management expectancy) is not violated in this path, and it seems it is optimum from some quantitative indexes.

MPLS since self route separate with forwarding, be fit to TE combine formation MPLS-TE technology.Use MPLS-TE, can improve the QoS of network, be mainly reflected in:

(1) utilizes MPLS-TE, can in many possible forward-paths, carry out load balance, thereby avoid congested, improve QoS.

(2) use MPLS-TE, create one by the RSVP-TE signaling and have the tunnel that strict QoS bandwidth guarantees, thereby support absolute QoS.

(3) can carry out Additional Protection to the tunnel by backup LSP, FRR modes such as (heavy-routes fast), thereby improve the QoS of network.

But the limitation of MPLS-TE is still obvious, comprising: at first it must be applied in the MPLS network, and therefore the non-MPLS existing network of part can't be supported this The Application of Technology at present; Secondly at present to the cross-domain application of MPLS-TE still in conceptual phase, this means that the main application of current MPLS-TE can only be in individual domain; In addition, create and to have the tunnel that bandwidth guarantees though MPLS-TE can be the user, if when transmitting multiple business simultaneously in the tunnel, how the business of these different priorities being carried out differentiated treatment also is the problem that needs research.

＜6〉bandwidth broker

In order more effectively to monitor and control the resource of the whole network, in the model of a new generation, people have proposed bandwidth broker (BB, Bandwidth Broker), the just notion of network resource manager again.The topological sum node and the link-state information of bandwidth broker collection network, network resource administration, and admit control in conjunction with the strategy of strategic server regulation.Simultaneously, Bandwidth Management also is in charge of the communication between the territory, reach by communicating by letter with the bandwidth broker of adjacent network domain cross-domain between the purpose that realizes of QoS.

The advantage of bandwidth management techniques is that QoS key-course and data transfer layer are separated, and core router does not need to carry out access control and the QoS state information is preserved, and has also simplified between the router and has eliminated qos signaling in other words, has simplified the complexity of router.This in addition mode supports absolute QoS to guarantee, comprises supporting cross-domain QoS to guarantee.Be exactly control uniformly of Internet resources quilt and management in this mode in addition, help telecom operators QoS is carried out as a kind of business.

But bandwidth management techniques still is in conceptual phase at present, the Signalling exchange between BB and operation layer and the bearing layer equipment, and the Signalling exchange between the BB is all also among discussing.In addition, this mode is very high to the requirement of BB, in some cases, if apply for that simultaneously the Business Stream number of resource is a lot, might make BB become bottleneck in the network.

4.PWE3

PWE3 originally was called Martigny draft (Martini Draft), and its original intention is an emulation Ethernet on MPLS.PWE3 is a kind of mechanism at the various point to point service, PTP of the last simulation of packet switching network (PSN), and the business that simulated can be E1 special line, ATM, FR or Ethernet etc.PWE3 utilizes the tunneling mechanism on the PSN to simulate a kind of necessary attribute of business, and the tunnel here is called dummy line (PW).PWE3 can encapsulate the protocol Data Unit (PDU) of special services, and PDU contains necessary data of emulation special services and control information in the inside.Use PWE3 mechanism, operator can transfer to all transport services among the network (as IP/MPLS) of a fusion.From user's angle, can think that the dummy line of PWE3 simulation is a kind of link or circuit of special use.

The PWE3 task groups has the standard that goes up transmission E1 aspect at PSN (Packet Switch Network, packet switching network), and the frame structure that PSN is gone up transmission E1 has had complete definition.

Fig. 2 is the functional schematic that PSN goes up transmission E1.

CE (Customer Edge, customer edge): it is to initiate or a side that termination is professional, and CE not can be appreciated that or not self adopt is the mode of emulation mode rather than original business.

PE (Provider Edge, provider's edge device): it provides PWE3 business for CE.

PW (PseudoWire, pseudo-line): be based on the connection between two PE equipment on the packet network.PE provides the adaption function between CE and the PW.

PSN (packet switching network) tunnel: can transmit a plurality of pseudo-lines (PW), make it transparent by packet switching network (PSN).

Two PE (PE1 and PE2) equipment makes between the CE and can communicate by letter by PSN for the CE equipment (CE1 and CE2) that they are connected respectively provides one or more pseudo-line.PSN tunnel is used for providing a data channel to pseudo-line.The business of pseudo-line is sightless for core network, and promptly core network is transparent for CE.

Fig. 3 then is that its protocol hierarchy structure is represented.

Wherein encapsulated layer provides necessary structure for the given load type that transmits on pseudo-line.The PW encapsulated layer comprises three sublayers: the load converge, regularly, the ordering.

PWE3 also goes up transmission E1 to PSN the adaptive capacity of PSN performance has been proposed demand:

Packet loss: the emulation of E1 circuit edge-to-edge can suppose that the end-to-end packet loss of Packet Based Network is lower, and is special, do not need retransmission mechanism.For minimum is reduced in the influence of Packet Based Network packet loss, encapsulated layer should possess following function:

(1), should be able to independently resolve the E1 data of each packet encapsulation in the outlet of E1.

(2) packet lost of reliable detection.

(3) packet is reduced to minimum to the influence of clock recovery.

(4) elasticity of increase interface, data that can be suitable replace the packet of losing.

Packet sequence: encapsulated layer must guarantee to carry the succession of the data packet delivery of E1 information:

(1) can detect out of order state.

(2) can change in proper order again.

Congested control: the E1 circuit has the constant bit rate, and the load that brings to PSN also is constant.Need be such as the flow-control mechanism of TCP.

5.ITU-T 8261

The description that ITU-T 8261 is detailed Packet Based Network transmit E1 synchronously and the regularly requirement of aspect.

ITU-T 8261 points out, except same reference clock was provided at two ends, the clock recovery when Packet Based Network transmits E1 mainly contained two kinds of schemes:

＜1〉differential clocks restoring method (Differential Methods).

Two ends can both provide reference clock, and can measure the deviation of local terminal clock and reference clock, and this deviation is sent to far-end, and far-end obtains the E1 clock according to this difference in conjunction with reference clock.

＜2〉adaptive clock recovery method (Adaptive Methods)

Two ends can not provide reference clock simultaneously, and business clock is according to data packet recovery.

The regulation of the drift of 8261 pairs of E1 interfaces of ITU-T is as shown in the table:

Observe τ (second) at interval	MRTIE requires (us)
		0.05＜τ≤0.2	40τ
0.2＜τ≤32	8
		32＜τ≤64	0.25τ
64＜τ≤1000(Note)	16

ITU-T 8261 points out that the information errors that packet loss causes is very big to the influence that Packet Based Network transmits E1.One of reason is that in the grouping make mistakes in a certain position, will cause whole data-bag lost, thereby cause the unexpected error of E1 information flow.Therefore moderate packet loss (with reference to the situation of conventional Packet Based Network) will cause the E1 circuit unavailable.

6.FEC technology

FEC is a kind of error correcting technique, in communication and field of storage application is more widely arranged.

The basic principle in communications applications of FEC is: for the information digit certificate that will transmit the k bit, according to certain coding method it is encoded, generate the n (data of bit of n＞k), the redundant digit of the n-k bit of the information digit certificate of k bit and generation is sent together, and the recipient can utilize redundant digit to recover the information digit certificate of losing in the transmission; The recipient with the k ' that receives (k ' 〉=k) data of bit, use corresponding interpretation method to recover the information digit certificate of k bit.

The FEC technology can be expressed as in the basic principle of the application of field of storage: before data are being stored in storage medium, carry out encoding process by certain algorithm in advance, add the redundant code that has the feature of data own, from the storage medium reading of data time, according to respective algorithms the data of reading are decoded, thereby find out error code that storage medium produces and its correction.

Based on bit-level, the FEC mechanism of byte level and data block has obtained to use more widely.

Also adopted packet-based FEC technology in field of video processing.

7. Packet Based Network transmits the deficiency of E1

Packet Based Network transmits the existing actual application of E1.

Packet Based Network no matter be IP or Ethernet, all is based on the asynchronous connectionless network of statistic multiplexing, a little less than the support relatively to QOS (Quality Of Service, service quality).

Above-mentioned various schemes, major part can not realize fully that QOS guarantees.Adopt IntServ or MPLS technology can farthest guarantee QOS, but be not that conventional network equipment can provide, and implement complexity and cost is all higher.Simultaneously, no matter which kind of scheme all is to need the all-network node device in the packet network that corresponding support is provided, and this is difficult to guarantee in existing packet network.In other words, conventional network equipment can not all be supported a kind of in these schemes.This makes that the QOS of transmission E1 can not adopt these schemes to ensure on existing packet network.

Therefore the QOS ability of Packet Based Network is with very outstanding to the contradiction of the high request of QOS based on synchronous E1.When transmitting E1 with packet network, packet loss that network congestion causes or time delay surpass thresholding all can show as losing of information at receiving terminal, and this has seriously influenced the quality of E1.

Existing Packet Based Network transmits technology and the product of E1, relies on the assurance that Packet Based Network provides QOS more.Can not ensure the performance of its requirement and when causing information dropout, how replace the data of losing at packet network, or come the frame of place of lost with the previous frame of receiving with complete " 1 ".This obviously can't guarantee the quality of E1.

Do not transmit the precedent that E1 uses FEC though have, by theory analysis as can be known, adopt FEC mechanism on existing packet network, to transmit E1, can obtain to improve greatly in performance in Packet Based Network.

Summary of the invention

TDM is very high to the requirement of QOS, and the QOS system of existing Packet Based Network technology is not perfect, can not provide QOS to guarantee.

Standard that the PWE3 task groups proposes and draft are not also stipulated and are required Packet Based Network to provide QOS to guarantee.

Need rely on packet network just because of the technology of existing Packet Based Network transmission TDM and product provides QOS to guarantee, and the node device in the existing packet network can not provide QOS to ensure.So Packet Based Network transmission E1 need make it needn't rely on the QOS system of packet network, thereby expand the application of self in conjunction with other technology and mechanism.

Also depend on the problem of the performance of packet network unduly in order to solve TDM business such as Packet Based Network transmission E1 just, the present invention proposes the method for a kind of FEC of employing mechanism reliable end-to-end E1 of transmission on Packet Based Network.Its objective is: under the relatively poor situation of network QOS, still can ensure the end to end performance that E1 transmits on Packet Based Network by FEC mechanism.

Core content of the present invention can be described with following three steps:

A) (E1 enters an end of Packet Based Network at the source node, also can be described as the transmission node, distinguish source node and destination node only for sake of convenience, only one-way transmission relatively, source node all has transmission/receiving function simultaneously with destination node) E1 is become a data packet sequence according to specific rule with protocol encapsulation, and adopt specific coding rule and agreement to generate a FEC data packet group according to this sequence of data packet.

B) this sequence of data packet and FEC data packet group are sent to destination node (E1 leaves an end of Packet Based Network, also can be described as to receive node) by Packet Based Network.

C) in described destination node according to FEC package and sequence of data packet, as have data-bag lost, then the packet of losing according to specific rule regeneration extracts or the regeneration E1 net load and the E1 code stream that reduces.

The present invention can compatible not contain the similar technology of FEC mechanism, also, have FEC mechanism the source node can with the destination node interconnection that does not have FEC mechanism.

The present invention determines in the mode of auto-negotiation whether destination node possesses identical FEC mechanism.Do not possess identical FEC mechanism as destination node, then local terminal is arranged to non-FEC mode of operation.

The encapsulated layer of Packet Based Network of the present invention has adopted Real-time Transport Protocol, as regularly (particularly differential clocks reset mode) recovery.Also be consistent simultaneously with the standard that IETF PWE3 task groups is formulated.Do not relying on the application that RTP realizes that timing recovers, this agreement can be used as reservation function.

When being configured to the differential clocks reset mode, the time stamp field of Real-time Transport Protocol is carried the difference information of local clock and reference clock.

When being configured to the adaptive clock recovery mode, the time stamp field of Real-time Transport Protocol is carried the time stab information that generates according to the local service clock.

The FORMID of the control word of the encapsulated layer of Packet Based Network of the present invention adopts and expands definition, and in order to the machine-processed rule of using of sign FEC.The present invention utilizes the rule that leftmost reserved field (totally four bits) definition is adopted with sign FEC mechanism in the control word (TDMoIP) of the encapsulated layer of Packet Based Network.The present invention redefines back called after FEC type with this reserved field.

The length field of the control word of the encapsulated layer of Packet Based Network of the present invention is not used further to represent length information, but arrives local unidirectional end-to-end packet loss in order to the far-end of sign local measurement, and inferior information is fed back to far-end by packet.The present invention is with length field called after length/performance.

The sequence number of the control word of the encapsulated layer of Packet Based Network of the present invention is defined as a two-dimensional structure, chief series number and subsequence two-stage.Chief series is number corresponding with the time slot group, and in order to identifying a data package, subsequence is number then in order to distinguish each packet in the data packet group.Wherein sequence number 1-k uses for the time slot group, and sequence number k-n is that the FEC group is used.

Packet switching stratum reticulare of the present invention adopts IP agreement and frame structure.

The data link layer of Packet Based Network of the present invention adopts ethernet link layer protocol and frame structure.

A) step of the present invention also further comprises following substep:

The present invention designs 32 business clock counters to the local service clock count.Designing 32 reference clock counters simultaneously counts outside reference clock.

During the mapping (enum) data bag, the E1 code stream is divided into groups to press time slot.After a complete time slot entered buffering area, the value of record traffic clock counter and reference clock counter was to be used to generate the time stamp of this time slot.

During the mapping (enum) data bag, the RTP field need be carried definite time stab information in all packets that the mapping of time slot group generates.

If be configured to the self-adaptation clock pattern, then time stab information generates according to the business clock count value of record.

If be configured to the differential clocks reset mode, then time stab information is according to the business clock count value and the external reference clock count value of record, and generates according to certain rule.

One group of E1 time slot (being called E1 time slot group) concentrated be mapped to one group of packet.Wherein the number of the E1 time slot that comprises in the E1 time slot group can be configured to 8,16 or 32.Why selecting the member number of 2^m as the time slot group, is because adopt such value, and design buffering area structure, respective handling and generation FEC error correction packets are all more convenient.

E1 time slot group is concentrated when being mapped to packet data package, can be disposed and use each time slot independently to be mapped to the independent mapping mode of bag, also can select each the time slot selected part data from E1 time slot group, be packaged into the synthetic mapping mode of packet again.

Independent mapping mode is handled comparatively simple, and it is fewer to be implicative of each other between the packet, but Information Security is relatively poor, and Packet Based Network can not provide the perfect assurance of information security.

Synthetic mapping mode is just in time opposite, can upset the arrangement of E1 payload information, is equivalent to information is encrypted, and helps the transmission (with respect to E1 network, the fail safe of Packet Based Network relatively poor) of E1 information at the grouping internet safe.But it handles comparatively complicated, and at that rate, and the data coupling that E1 time slot group comprises is very tight, or whole mistakes, or entirely true.

One group of E1 time slot is concentrated when being mapped to packet data package,, then selected corresponding position to be combined into byte or word, recomposition frame from each byte of each time slot if use synthetic mapping mode.Comprising 8 time slots with the time slot group is example, with b (i, j, k) the k position of j byte of i time slot of expression, j byte of the clean information of i the packet that then is mapped to (with B (i, j) expression) can be expressed as b (1, j, i) b (2, j, i) b (3, j, i) b (4, j, i) b (5, j, i) b (6, j, i) b (7, j, i) b (8, j, i).

With an E1 time slot group concentrate be mapped to packet after, generate the FEC data packet group according to the grouped data packet sequence that generates.The data of FEC data packet group generate according to time slot group information, and the generative process of data packet head and general data bag are similar.

Generate the FEC package, adopt Reed-Solomon (n, k) coding.This coding scheme is the non-binary code system, is that unit realizes EDC error detection and correction with the data block, and n is the length of data block, and k is the clean message length that data block comprises, and n-k is an error correction code length.

Generate the FEC package, and employing Reed-Solomon (n, when k) encoding, k can be configured to 8,16 or 32, and the value of n can have options such as k+4, k+8, k+16, k+32.These options can dispose automatically according to the performance information of far-end feedback.

The step that generates the FEC package is:

The i combination of bytes of each time slot (information that comprises 32 bytes) in k the time slot (that is to say that a time slot group comprises k time slot) is obtained the k byte sequence together, according to this k byte sequence, and generate the error correcting code of n-k byte in conjunction with generator polynomial, this n-k byte is respectively as the 1st, the i byte of 2...n-k FEC bag.

B of the present invention) step also further comprises following substep:

When sending packet, the transmitting time of general data bag depends on packet (or its time slot that carries) corresponding service clock count value, that is to say that the difference of the transmission time interval of two adjacent data bags and two adjacent business clock count values is proportional in the time slot group.

When sending packet, in the transmission at interval of two general data bags, send the FEC packet.The FEC packet number that each time interval sends is determined according to the value of n and k.The time that sends the FEC packet determines that principle is: the time interval that the adjacent data bag is sent is even.

C of the present invention) step also further comprises following substep:

For the control time delay reception timer is set at receiving terminal, the cycle of overflowing equals k*125us.

After the packet that receives removed frame head, the bag sequence number according to packet carries it is believed that with the net amount that obtains breath deposits the relevant position of jitter-removing buffer in.

If be configured to the self-adaptive recovery pattern based on time stamp, the time stab information in the Real-time Transport Protocol that then needs packet is carried deposits the far-end time stamp memory space of buffering area in.Utilize the local service clock to generate time stamp simultaneously, and this time stamp is deposited in the local time stamp memory space of buffering area.

Recover pattern if be configured to differential clocks, then obtain the local service clock in conjunction with external clock with this time stamp.

If be configured to self-adaptive recovery pattern based on time stamp, and the not storage as yet of the time stab information of this time slot group, then need to deposit the time stab information in the Real-time Transport Protocol in the time stamp buffering area.

After the FEC that receives bag removed packet header,, the clean error correction code information that obtains is deposited in the relevant position of error correction buffering area according to the bag sequence number that packet carries.

When the reception timer overflows, check the data of jitter-removing buffer, as have information dropout, then utilize the information of error correction buffering area, in conjunction with the data of jitter-removing buffer, the clean information that the packet that regeneration is lost carries.

When receiving timer and overflow, and after drop-out recovers to finish, need to recover the business clock of E1.Can according to whether having reference clock, the restoration methods of configuration E1 clock is differences method and adaptive approach.

Differences method is the time stamp that utilizes Real-time Transport Protocol to carry, and generates the local service clock in conjunction with the external reference clock.

Adopting adaptive approach, is operating position or the time stab information according to jitter-removing buffer, adopts the mode recovered clock of feedback regulation.

Adopt adaptive approach, can select for use as required based on jitter-removing buffer filling level with based on two kinds of patterns of RTP time stamp.

Adopt adaptive approach, the accuracy rating of business clock that can configuration restore.

If when being configured to the adaptive mode recovered clock, and be configured to use when cushioning the filling level, then recover the local service clock according to the operating position of buffering area.

If when being configured to the adaptive mode recovered clock, and when being configured to use time stamp, then the time stab information according to buffer stores recovers the local service clock.

Then, use the business clock that recovers, and the data in binding buffer district are recovered the E1 code stream.

The present invention also provides a kind of equipment of above-mentioned method to realize on Packet Based Network that by FEC mechanism reliable end-to-end E1 transmits of using.

Description of drawings

Fig. 1 represents major function flow process of the present invention (describing the professional transmission of unidirectional E1).

Fig. 2 is the function signal of IETF PWE3 task groups about the circuit simulation of edge-to-edge.

Fig. 3 is the protocol hierarchy structure of IETF PWE3 task groups about the circuit simulation of edge-to-edge.

Fig. 4 be IETF PWE3 task groups about the protocol architecture on the circuit simulation PSN layer IP agreement of edge-to-edge, also be the protocol architecture that the present invention adopts.

Fig. 5 is the standard of the relevant TDMoIP of IETF PWE3 and the diagram of the control word that draft defines.

Fig. 6 is the diagram of the control word that defines of the present invention.

Fig. 7 is the hierarchy diagram of the agreement used of the present invention.

Fig. 8 is the frame structure diagram of the agreement used of the present invention.

Fig. 9 is the functional structure chart of the equipment of invention.

The signal that Figure 10 employing of the present invention FEC mechanism transmits E1.

Figure 11 functional module detailed annotation of the present invention.

The diagram that Figure 12 FEC bag of the present invention generates.

Figure 13 represents to utilize FEC mechanism to correct a mistake to the situation of the improvement of physical channel performance.

The flow chart of Figure 14 coding module of the present invention.

The flow chart of Figure 15 package module of the present invention.

The flow chart of Figure 16 parse module of the present invention.

The flow chart of Figure 17 decoder module of the present invention.

Figure 18 flow chart of filling the adaptive clock recovery module of level pattern based on buffering area of the present invention.

The schematic diagram of Figure 19 jitter-removing buffer of the present invention.

The flow chart of Figure 20 adaptive clock recovery module based on RTP time stamp pattern of the present invention.

The front elevation of the equipment of Figure 21 invention.

The back view of the equipment of Figure 22 invention.

Figure 23 Performance Evaluation of the present invention and test network topology.

Embodiment

Core content of the present invention is realized with FPGA, and is used in the equipment of invention.

Fig. 1 represents major function flow process of the present invention.

101 expression transmitting terminals are mapped to E1 the process of buffering area.

102 expression transmitting terminals become data encapsulation the process of packet.

The process that 103 expression transmitting terminal FEC packets generate.

The data that 104 expression receiving terminals are received deposit the process of buffering area in.

105 expression receiving terminals utilize the process of the mechanically manufactured regenerated packet of losing of FEC.

After 106 expression receiving terminal FEC mechanism are handled, utilize the process of reduction of data E1 code stream.

107 expression Packet Based Network.

The step of implementing is respectively:

1) definite frame format that adopts.

2) determine the coding rule that FEC mechanism is used.

3) time stamp buffering area design

4) send the buffering area design.

5) jitter-removing buffer design.

6) error correction buffering area design.

7) receive timer design.

8) package modular design.

9) parse module design.

10) coding module design.

11) decoder module design.

12) clock recovery module design etc.

13) the opposite end type is determined (whether possessing FEC mechanism).

It below is the detailed content of each step

1) definite frame format that adopts.

Fig. 3 has described the frame structure of the standard declaration of IETF PWE3 task groups formulation, and the present invention is a template with this frame structure.But owing to will adopt FEC mechanism, and transmitting terminal should be able to be by the configuration of frame head information announcement receiving terminal FEC mechanism, so that the correct decoding of receiving terminal, so that frame structure can not be with standard declaration is just the same.

Be the consideration of interconnecting property simultaneously, standard and draft compatibility that frame structure of the present invention and IETF PWE3 task groups are formulated.

Be conformance with standard and compatible consideration, the present invention does not change the main framework of aforementioned frame format, mainly contains the change of the following aspects:

(1) redefine the FEC type field with IETF PWE3 about the reserved field of four bits of the Far Left in the TDMoIP control word, the FEC type is represented the configuration information of FEC mechanism.

(2) redefine the FEC enable field with IETF PWE3 about the reserved field of the 8th, 9 two bit in the TDMoIP control word, FEC enables expression and whether forbids FEC mechanism of the present invention.

(3) 16 bit sequence number are decomposed into the structure of a two dimension: high 10 as chief series number, chief series identification data package, low 6 as subsequence number, each packet sequence in the expression data packet group.

(4) length field redefines and is length/performance level field, is used for representing the unidirectional end to end performance of network.

Former protocol architecture as shown in Figure 5, each field is respectively:

RES (four of Far Lefts), reserved field.When indication was the MPLS network, these four should be provided with and help 0.

The L field, 1, indication is local lost efficacy.

R field, 1, indicating remote lost efficacy.

The M field, 2, the implication of indication L field, current undefined.

RES (the 8th and the 9th), reserved field should be set to complete 0.

Length field, 6, indication comprises the length of control word at interior net load.

Sequence number is used for the sequencing of packet.

The protocol architecture of the present invention's definition as shown in Figure 6.Each field is respectively:

The FEC type, 4.Specifically define referring to following description.

The L field is 1, identical with former definition.

R field is 1, identical with former definition.

The M field, 2, the implication of indication L field, current undefined.

FEC mechanism enables (the 8th and the 9th), is set to 11 expressions FEC mechanism of the present invention and enables, be set to represent in 00 o'clock, for the network equipment intercommunication of the non-FEC mechanism of far-end, forbid FEC mechanism of the present invention.

Performance level, 6, the unidirectional end to end performance of expression network.Have 64 values, be defined as follows:

The unidirectional no packet loss of 0---.

The unidirectional packet loss of 1---is less than or equal to 1%.

The unidirectional packet loss of 2---is greater than 1%., but be less than or equal to 5%.

The unidirectional packet loss of 3---is greater than 5%., but be less than or equal to 10%.

The unidirectional packet loss of 4---is greater than 10%., but be less than or equal to 15%.

The unidirectional packet loss of 5---is greater than 15%., but be less than or equal to 20%.

The unidirectional packet loss of 6---is greater than 20%., but be less than or equal to 25%.

The unidirectional packet loss of 7---is greater than 25%., but be less than or equal to 30%.

The unidirectional packet loss of 8---is greater than 30%., but be less than or equal to 35%.

The unidirectional packet loss of 9---is greater than 35%., but be less than or equal to 40%.

The unidirectional packet loss of 10---is greater than 40%., but be less than or equal to 45%.

The unidirectional packet loss of 11---is greater than 45%., but be less than or equal to 50%.

The unidirectional packet loss of 12---is greater than 50%., but be less than or equal to 55%.

The unidirectional packet loss of 13---is greater than 55%., but be less than or equal to 60%.

The unidirectional packet loss of 14---is greater than 60%., but be less than or equal to 65%.

The unidirectional packet loss of 15---is greater than 65%., but be less than or equal to 70%.

The unidirectional packet loss of 16---is greater than 70%., but be less than or equal to 75%.

The unidirectional packet loss of 17---is greater than 75%., but be less than or equal to 80%.

The unidirectional packet loss of 18---is greater than 80%., but be less than or equal to 85%.

The unidirectional packet loss of 19---is greater than 85%., but be less than or equal to 90%.

The unidirectional packet loss of 20---is greater than 90%., but be less than or equal to 95%.

The unidirectional packet loss of 21---is greater than 95%., but be less than or equal to 100%.

22-63---keeps.

Chief series number, 10, the group number of expression data packet group.

Subsequence number, is represented the order of each packet in data packet group by 6.

Fig. 7 is that the present invention adopts complete frame structure.

The 701st, Ethernet frame head, totally 14 bytes.

The 702nd, IP head, totally 20 bytes.

The 703rd, the UDP head is 8 bytes.

The 704th, Real-time Transport Protocol, totally 12 bytes.

The 705th, carry the control word of FEC mechanism, be 4 bytes.

The 706th, TDM net load, totally 32 bytes.

2) determine the coding rule that FEC mechanism is used.

FEC usually is used for correcting the error code of physical communication and memory device, uses very extensive.The mainstream technology SDH system of transmission network has also adopted FEC mechanism.

Newly do not find that Packet Based Network transmits the precedent that E1 adopts FEC mechanism but look into.

Here adopt R-S (n, k) coding scheme.

Theory analysis shows that (hamming (Hamming) distance is for n, k) coding to adopt R-S

d _min＝n-k+1

The wrong number that then can correct is at most:

And the deletion error correcting capability is:

ρ＝d _min-1＝n-k

The generator polynomial that the present invention uses is:

G(x)＝x ⁸+x ⁴+x ³+x ²+1

Recover the packet loss that Packet Based Network transmits E1, just each byte of the packet of losing by FEC regeneration.Can judge by sequence number and the position of the grouping of losing also just can determine the position that each data block is made mistakes, so its essence is a kind of deletion error correcting capability.Therefore, adopt R-S (n, k) coding, when the number of each group data-bag lost during less than (n-k), all right-on generations of available decoder of the packet of losing.

The value of k of the present invention has 8,16,32 3 options.

When k=8, n can select four values such as 12,16,24,32 grades for use.

When k=16, n can select four values such as 20,24,28,32 grades for use.

When k=32, n can select four values such as 40,48,56,64 grades for use.

The value of k can be by software arrangements, and the value of n then has the feedback negotiation mechanism to determine, that is to say, when the packet loss of measuring when far-end was low, the value of n can be arranged to less number, otherwise then can be arranged to bigger number.

For example, when k=8, the packet loss that far-end is measured was more than or equal to 20% o'clock, and the value of n chooses 32, and the packet loss of measuring when far-end is when equaling 10%, and the value of n chooses 16.

In conjunction with aforesaid frame structure, if represent described FEC type, then can represent to have the packet of FEC mechanism with the value of b3b2b1b0 with b3b2b1b0, be defined as follows:

B3b2:00---represents not adopt FEC mechanism, adopts when packet loss remains at low levels.

01---represents to adopt FEC mechanism, and k=8,

10---represents to adopt FEC mechanism, and k=16,

11---represents to adopt FEC mechanism, and k=32.

B1b0:00---when k=8, n=12,

When k=16, n=20,

When k=32, n=40

B1b0:01---when k=8, n=16,

When k=16, n=24,

When k=32, n=48

B1b0:10---when k=8, n=24,

When k=16, n=28,

When k=32, n=56

B1b0:11---when k=8, n=32,

When k=16, n=32,

When k=32, n=64

When the FEC enable field is 00, indication forbidding FEC field, the definition of b3b2b1b0 was at this moment lost efficacy, and should be set to complete 0.

Be example with k=8 below, analyze the performance performance of FEC mechanism of the present invention on existing network.Each packet in transmission over networks is actually relatively independent, there is no contact each other.Now suppose on packet network because the congested actual packet loss that causes is p, according to the transmission network performance evaluation as can be known, transmit n data packet loss m (probability P (m) of the packet more than m＜n) is individual is deferred to binomial distribution, can represent with following formula:

$P\left(m\right)=\underset{j=m}{\overset{n}{\mathrm{\&Sigma;}}}{C}_n^j{p}^j{(1-p)}^{(n-j)}$

So, when m＜=n-k, the packet of losing can utilize paid-in packet and the holomorphosis of FEC packet.Otherwise, then can not recover.

Following table is the relation of P (k=8) and p, and P (8) is when k=8, when end-to-end packet loss is p, and the minimum packet loss that adopts FEC mechanism to reach.

p(％)	5	10	15	20	30	40	50
								P(8)(ppm) n＝16	0.01	58	140	1866	18540	83950	174560
P(8)(ppm) n＝32	0	0	0	0	0.02	120	1000

This shows:

When adopting R-S (32,8) coding and realize FEC mechanism, though network congestion up to 50% o'clock, the probability of real data packet loss is also only 0.1%, such packet loss can satisfy the requirement of application.

3) time stamp buffering area design

The time stamp buffering area should be able to be stored three kinds of time stamps, and is as follows respectively:

(1) the TDM time slot enters the time stamp (value of business clock counter) of buffering area.

(2) time stamp that the packet of receiving carries (coming from RTP time stamp territory).

Local time stamp (value of business clock counter) when (3) receiving packet.

Time stamp buffering area of the present invention is designed to 1024 bytes, can store 256 time stab informations altogether.

4) send the buffering area design

The transmission buffering area here is meant that when E1 information enters Packet Based Network buffer memory E1 information is so that encapsulation forms the buffering area of packet.Need the information spinner of buffer memory that following three parts will be arranged:

(1) buffer memory of time slot group: owing to be that unit encapsulates with the time slot group here, so buffer size can not be less than the size of a time slot group.Here a time slot group maximum can comprise 32 time slots, and the length of the net amount certificate that each E1 time slot comprises is 32 bytes, and therefore, the size that sends buffering area should be greater than 32*32=1024 byte=8192 bits.

(2) buffer memory of FEC information: simultaneously, after time slot group of buffer memory, need to generate FEC information, and time slot group and FEC information all are packaged into packet, therefore the space of buffer memory FEC information and packaged packet also need be arranged.

By aforementioned content as can be known, FEC information is not more than the message length that maximum time slot group (32 time slots) comprises.

(3) buffer memory of packet: simultaneously the quantity S of cached data packet can be smaller, even but S is an also operate as normal under 1 the situation, but S is 1 o'clock, the loss on the efficient is very big, deteriorates to the work in series pattern that afterwards sends that encapsulates earlier substantially.

Here choose 4 quantity as cached data packet, both can accomplish the package process and send process data packet is parallel work-flow, and the buffer memory quantity that takies is also less, the size of total data cached bag＜512 bytes.

The total amount that sends buffering area is designed to the 3K byte, both meets the demands, and certain spare space storage time stab information is arranged again, or supplies the need of functions expanding.

When sending packet, the transmitting time of general data bag depends on packet (or its time slot that carries) corresponding service clock count value, that is to say that the difference of the transmission time interval of two adjacent data bags and two adjacent business clock count values is proportional in the time slot group.The FEC packet sends in the transmission at interval of two general data bags, and wherein the FEC packet number of each time interval transmission is determined according to the value of n and k.The time that sends the FEC packet determines that principle is: the time interval that the adjacent data bag is sent is even.

5) design of jitter-removing buffer

Jitter-removing buffer of the present invention has two functions: under the situation that does not have regularly transfer mechanism, the timing information that is used for adaptive mode recovers and is used for the regeneration of drop-out.

Regeneration as drop-out, then the size of jitter-removing buffer can be deposited the amount of information of an E1 time slot group correspondence at least, since k＜=32, and the time slot of one road E1 is 32 bytes, and then the minimum size of jitter-removing buffer is 32*32=1024 byte=8192 bits.The size of actual jitter-removing buffer is arranged to L times of minimum size, L=2, and 4,8... can decide according to the capacity of the programming device (being FPGA here) that uses.

As the recovery of self-adaptation clock, then the size of jitter-removing buffer is the bigger the better in theory, and can not be too little.Because jitter-removing buffer is too little, still during non-locking, overflow and underflow can appear in jitter-removing buffer, cause loss of data or stationary problem in clock recovered.

On the other hand, jitter-removing buffer is excessive, and the time delay that then causes has bigger increase.

The structure of the jitter-removing buffer that Figure 19 represents is a loop buffer pool structure.

1901 these buffering area time slot corresponding group data of expression arrive fully or partly arrive but recover fully by FEC mechanism, do not revert to the E1 code stream but still be used as.

This cushions that trivial time slot corresponding group free of data bag arrives or data of this buffering area have been used as and revert to the E1 code stream 1902 expressions.

The data division that 1903 these buffering area time slot corresponding groups of expression comprise arrives (all not arriving), and does not use FEC to recover the situation of obliterated data.

1901 these buffering area time slot corresponding group data of expression arrive fully or partly arrive but recover fully by FEC mechanism, and are being resumed the situation of E1 code stream.

The packet (dash area) that has arrived in 1905 expressions, the 1903 time slot corresponding groups.

The packet (blank parts) of no show in 1906 expressions, the 1903 time slot corresponding groups.

The write pointer of the annular Buffer Pool of 1907 expressions, last byte of the information that the packet of the subsequence maximum that has arrived in the sensing current time slots group comprises.

The read pointer of the annular Buffer Pool of 1908 expressions points to the data that will read when recovering the E1 code stream.

General, when adopting the adaptive approach recovered clock, the initial level of filling is about 1/2, so the time delay that jitter-removing buffer causes is about the time of handling half jitter-removing buffer.If jitter-removing buffer can be stored L time slot group, then jitter-removing buffer can bring the time delay of L/2*k*125us.

E1 has the higher requirement of ratio to time delay.E1 carries voice messaging more, and the time delay of end-to-end speech information reaches 25ms when above, will produce apparent in view echo, and voice quality is descended.Though use special echo cancelling device can suppress echo, cost is higher.Therefore, the design jitter-removing buffer can not cause time delay excessive.

The end-to-end time delay D that Packet Based Network transmits E1 is transmitting terminal processing delay Dht, transfer delay Dt, transmit time delay D f reached receiving terminal processing delay Dhr accumulation and.

The long Distance Transmission of existing network generally uses optical fiber as transmission medium, and the propagation delay time that 100 kilometers optical fiber brings is about 300～500US, and therefore, in most cases, the influence of Dt can be ignored.

Existing apparatus for network node generally adopts store-and-forward mechanism, therefore can bring certain forwarding time delay, the forwarding time delay sum of the path all-network node device that the end-to-end forwarding time delay of whole group net should be a process.But the forwarding mechanism of existing network node device adopts high speed ASIC or NP to realize not having under the situation of network congestion more, and it is less relatively to transmit time delay.

If occurring under the congested situation, the then actual time delay that causes is the amount of a change at random, might be very big.In other words, delay variation might be in a high level.But use FEC mechanism in conjunction with overtime timer, bigger delay variation is inhibited.Time-delay when therefore analyzing can be used as congested situation the situation that does not have network congestion and handle.

What following table was represented is under no congested situation, the time delay that the packet of the different length of certain apparatus for network node actual measurement produces by this network equipment.

Transmit time delay (US)	4.74	7.7	11.2	17.7	22.3	26.7
							Data packet length (BYTE)	64	128	256	512	1024	1518

Because packet of the present invention all below 128 bytes, therefore, needs just can cause through this network equipment more than 120 time delay of 1ms, therefore, in most cases also can ignore the influence of Df.

Therefore, shorter in end-to-end distance, and less through the network equipment on the path, D ≈ Dht+Dhr then.

Owing to complete time slot group need be mapped to packet simultaneously at transmitting terminal, and the cycle of each time slot is 125us, therefore transmitting terminal can bring the buffer delay of intrinsic k*125us, jitter-removing buffer then also can bring the time delay of L/2*k*125us, add the package time delay, the coding time delay, decoding time delay and the summation of separating packet delay can not surpass thresholding, therefore L can not be provided with excessive under situation about meeting the demands.

Package time delay+coding time delay is relevant with the size of time slot group, if k is 32, then package time delay+coding time delay and can not be greater than 32*125us=4ms.

Equally, the decoding time delay is conciliate packet delay, and also the size with the time slot group is relevant,, if k is 32, the time delay of then decoding+separate packet delay with can not be also greater than 32*125us=4ms.

If be that satisfied overall delay end to end is less than 25ms, then

L/2*k*0.125+ decoding time delay+separate packet delay+package time delay+coding time delay＜25,

L*k＜272 then, Here it is needs the condition that satisfies.

The present invention designs the L value and is 32 to the maximum.

When k=8, the value of L generally is designed to 32.

When k=16, the value of L generally is designed to 16.

When k=32, the value of L generally is designed to 8.

6) receive timer design.

The function that receives timer is to provide timing information for data packet recovery.

Figure 19 represents jitter-removing buffer, and it is actual to be a circle queue, uses two pointer indications of read pointer P11 (shown in 1907) and write pointer P12 (shown in 1908).Its L that deposits time slot group can be regarded as orderly in logic, and chief series number relatively little (owing to be loop configuration, not having absolute order) in front.

Make S buffering area of write pointers point, read pointer points to T buffering area.Number with the FEC package of seasonal error correction buffer stores is R.

When the reception timer overflowed, system checked since (T+1) %L buffering area, finishes to S buffering area.

At first check the information of jitter-removing buffer, divide following two kinds of situation discussion:

＜1〉if (S+L-(T+1)) %L, represents the scope that this buffering area is not handled at this FEC greater than R, if this buffering area time slot corresponding group exists some or a plurality of time slots to lose, can not use FEC mechanism to recover, following three possibilities are then arranged:

If a last time slot of the time slot of losing exists, a then above time slot is filled this time slot, otherwise,

If the next time slot of the time slot of losing exists, then fill this time slot with next time slot, otherwise,

If a last time slot and the next time slot of the time slot of losing do not exist, then fill this time slot with complete 1.

＜2〉if (S+L-(T+1)) %L is less than or equal to R, represent the scope that this buffering area is handled at this FEC, if this buffering area time slot corresponding group exists some or a plurality of time slots to lose, whether then judge in conjunction with the information of error correction buffering area can error correction, as can error correction, then corresponding data is delivered decoder module, recover the information of losing by special algorithm.

Being provided with of the cycle of reception timer is extremely important.

Longly then can cause information dropout.If overflow excessive cycle, when the reception timer overflowed, possible read pointer had been crossed the time slot group corresponding buffer region that handle.Cause information dropout.

Cycle is too short may not guarantee in one-period, to handle all since (T+1) %L buffering area, all buffering areas that finish to S buffering area.

The duration that cycle approximates a time slot group is set here.

7) error correction buffering area design

The error correction buffering area is used to deposit the FEC packet, and by aforementioned content as can be known, the number-of-packet maximum that FEC data packet group comprises can reach 32, and therefore, the error correction buffering area must guarantee to deposit the data of 32 time slot correspondences.

When the network delay variable quantity is big, may cause a lot of buffering areas of jitter-removing buffer all to be in partially filled state (being that buffering area time slot corresponding group has only the Data Receiving of part time slot to arrive), opposite extreme situations, have L/2 buffering area and be in so situation, Dui Ying error correction buffering area maximum also should be able to be deposited L/2 FEC package so with it.

But determining does not by analysis need L/2 FEC package is set in most cases, and too big lower to the utilance of resource yet.

The size of the error correction buffering area of the present invention's design is for can deposit 8 FEC packages.

8) coding module design.

Figure 14 is the flow chart of coding module.

The packet that FEC mechanism of the present invention just is to use some redundant data packets to help regenerate and lose.(n, k) coding generates redundant data packets to adopt R-S.

(n, k) sign indicating number is a kind of nonbinary error correcting code to R-S, and the ability of very strong correction unexpected error is arranged.It utilizes a master data piece through computing, generates redundant block, redundant block is obtained one with the merging of master data piece have the data block of error correction.

Since the corresponding a plurality of general data bags of each time slot group, and each general data bag is not corresponding one by one with the FEC bag, therefore can not use encoder directly to obtain the FEC package.

A time slot group is concentrated when being mapped to packet data package, can be disposed and use each time slot independently to be mapped to the independent mapping mode of bag, also can select each the time slot selected part data from one group of E1 time slot, be packaged into the synthetic mapping mode of packet again

Main thought of the present invention is to utilize the corresponding byte (representing with i) of each time slot in the time slot group to be combined into the master data piece of a k byte, this data block obtains the redundant block of (n-k) length through codimg logic (for example remainder formation logic), and each byte of redundant block is respectively as the i byte of each packet in the FEC package.

If the end-to-end packet loss that is obtained by feedback information satisfies application requirements, then do not need to form the FEC bag.

9) package modular design

Fig. 4 is the protocol architecture that the present invention adopts.

Fig. 5 is the diagram of the control word of the TDMoIP that defines in IETF PWE3 task groups standard or the draft.

Fig. 6 is the diagram of the control word that defines of the present invention.

Fig. 7 is the complete protocol hierarchy structural diagrams of the present invention.

Fig. 8 is the complete frame structure diagram of the present invention.

The function of package module is exactly that the content that will send buffering area is packaged into Fig. 7, and 8 packets of describing send to receiving terminal by the Packet Based Network such as IP or Ethernet.Each packet payload is 32 bytes.The packet that is packaged into is 90 bytes: be respectively Ethernet head 14 bytes, and IP 20 bytes, UDP 8 bytes, Real-time Transport Protocol 12 bytes, control word 4 bytes, information payload 32 bytes.

Since introduce FEC mechanism, the res field in the control word of TDMoIP, and length field and sequence-number field all have new definition, referring to the content of aforementioned frame structure.

Equally, owing to being the unit package with the time slot group, the sequence number in the control word is defined as the structure of two dimension.Chief series numbers 10, subsequence numbers 6, preceding k of subsequence number is numbered the general data bag and uses, and the sequence number in k+1-n interval uses for the FEC packet.

Figure 15 is the schematic diagram of package module.

When generating the Ethernet frame head, what the source MAC of use and target MAC (Media Access Control) address should be with the physical device configurations is identical.

When generating IP packet header, what the source IP address of use and purpose IP address should be with the physical device configurations is identical.

The content of FEC type field is of crucial importance, can not make mistakes.

Sequence number is also extremely important, can not make mistakes.

10) parse module design

Figure 16 is the flow chart of parse module.

For simplified design is handled the time delay of bringing with reducing.This module is not resolved the content in Ethernet data bag frame head and IP packet header.Ethernet frame head and IP packet header only are used for realizing the packet forwarding on Packet Based Network.

At first analyze the FEC type field in the control word of packet, obtain the n and the k value of FEC mechanism.

Take out the sequence number in the control word of packet again, decompose and obtain chief series number and subsequence number.

The time slot group of chief series MOD jitter-removing buffer is counted L and is obtained the memory location of current time slots group at jitter-removing buffer or error correction buffering area.

As follows sequence number＞n then is illegal packet, need abandon.

As n＞=subsequence number＞k, then be the FEC packet, need deposit the relevant position of error correction buffering area in.

As follows sequence number＜=k then is the general data bag, need deposit the relevant position of jitter-removing buffer in.

The data that enter buffering area do not comprise all header packet informations.

11) decoder module design.

Figure 17 is the flow chart of decoder module.

Before decoding, need the value of paid-in packet of statistics and the receivable packet that arrives, and calculate end-to-end packet loss, so that feed back to far-end.

When the FORMID indication does not have FEC mechanism, do not need decoder module.

When checking that jitter-removing buffer is not found information dropout, decoder module is not worked yet.

Corresponding with coding module, decoder module essence can not directly be recovered the packet lost, but the information that the packet that byte-by-byte recovery is lost comprises.

Just, the subsequence of the corresponding byte of the packet received and FEC bag and lost data packets number is delivered decode logic,, deliver to the ad-hoc location of buffering area again by the byte that computing obtains losing.

12) clock recovery module design.

Do not using same external reference clock directly under the situation as the two ends business clock, clock recovery mainly contains two kinds of methods: differential clocks restoration methods and adaptive clock recovery method.

It still is the adaptive clock recovery method that the present invention can be defined as adopting the differential clocks restoration methods by configuration.

The requirement of differential clocks restoration methods has same reference clock at two ends.32 business clock counters of design design 32 reference clock counters simultaneously outside reference clock are counted the local service clock count among the present invention.

At transmitting terminal, after a complete time slot enters buffering area, write down the value of local business clock counter and external reference clock counter, and calculate the difference of local service clock and external reference clock, and this difference is sent to receiving terminal by packet (carrying in Real-time Transport Protocol time stamp territory)

At receiving terminal, from packet (Real-time Transport Protocol time stamp territory) extract transmitting terminal business clock and with the difference of external reference clock, and obtain the local service clock in conjunction with the external reference clock according to this difference.

The differential clocks restoration methods is comparatively direct, but needs to rely on same external reference clock, and narration here is omitted.

Here discuss the adaptive clock recovery module.

Because the delay problem of Packet Based Network, adaptive clock recovery method clock recovered has certain delay on phase place, but under lock-out state, frequency difference should be within the limits prescribed.

The basic principle of adaptive clock recovery is similar to the automatic feedback regulation mechanism of PLL.

Summarize to get up three major function pieces just:

" phase demodulation " logic: in fact this be not equal to the phase discriminator of PLL, but output on function with the output equivalence of phase discriminator.Just utilize the operating position of jitter-removing buffer, or utilize time stab information, the difference that obtains local service clock and far-end business clock is represented.

Loop filtering logic: use the filtering logic to eliminate the Packet Based Network time delay and change the shake that brings.

Voltage controlled oscillator/digital controlled oscillator: the output in conjunction with the loop filtering logic generates the local service clock.

The standard frequency of E1 is 2.048Mhz, and G.703 ITU-T is 2.048Mhz ± 50ppm to the requirement of E1 frequency accuracy.Therefore, the business clock frequency f of transmitting terminal _tShould drop on (2047897.6hz, 2048102.4hz) interval.Same receiving terminal is according to the clock frequency f of data packet recovery _rScope also should drop on this interval.

Ideally, f _r=f _t

But because the characteristic of Packet Based Network, in most cases, this equation can not satisfy.

Definition Δ f=f _r-f _t, from the above, Δ f must drop on (204.8hz, 204.8hz) interval, and MAX (| Δ f|)=409.6hz.

It is interval that (204.8hz 204.8hz) is exactly the scope of receiving terminal business clock around centre frequency (2.048Mhz) fluctuation.The purpose of self-adaptive recovery method will be eliminated this species diversity exactly, makes Δ f level off to 0 as far as possible.

When practical application, the adjusting range of receiving terminal business clock is configurable.

If transmitting terminal business clock f _tCorrupt to and exceed ITU-T regulation G.703, for making f _rFollow the tracks of f _t, the scope that can transfer big receiving terminal clock to adjust by configuration information.

Equally, if transmitting terminal business clock f _tHave higher precision, drift and jittering characteristic preferably, can turn the scope that the receiving terminal clock is adjusted down by configuration information.

Here represent the configuration information that clock is adjusted with w, its implication is that the precision of local service clock is 2.048Mhz+w ppm, and then the frequency range of local service clock is:

((2.048M-2.048w) hz, (2.048M+2.048w) hz), adjusting range is 4.096whz.

Self-adaptive recovery method of the present invention has two kinds of patterns: based on buffering area filling level pattern with based on RTP time stamp pattern.

(1) fills the level pattern based on buffering area

This pattern of use does not rely on far-end, and packet does not need the explicit timing information that carries.

Figure 18 is based on the functional schematic that buffering area is filled the adaptive clock recovery module of level pattern.

All do not revert to the byte number of E1 code stream and the ratio of whole jitter-removing buffer size in the definition filling level F=jitter-removing buffer.

The size of supposing whole jitter-removing buffer is a S1 byte, and then S1 can calculate with following formula:

S1＝L*k*32

As shown in figure 19, suppose that write pointer (1907) is P11, read pointer (1908) is P12.P11, P12 are the side-play amounts with respect to the logic original position of jitter-removing buffer.

Read pointer P12 can be byte-by-byte moving, and it is mobile by turn utilizing logic with shift even can accomplishing.Here P12 being regarded as byte-by-byte moves

Write pointer P11 is that unit moves with time slot (32 byte).It always points to the time slot group of up-to-date arrival in the jitter-removing buffer, and points to last byte of the time slot with maximum subsequence number that arrives in this time slot group.

Then F can calculate with following formula:

F＝(P11+S1-P12)/S1

Filling level side-play amount FP of the present invention is defined as the difference of F with respect to half occupied state (F=50%).That is:

FP＝F-0.5

Owing to constantly fill jitter-removing buffer from the packet that the Packet Based Network side is received, and the E1 side constantly sends the data of jitter-removing buffer, the operating position of buffering area is constantly to change like this.Therefore filling level increment FD is the function of time, along with time T changes.

Under the initial situation, F=0.5.

When the clock that extracts is faster than the business of transmitting terminal, then fill level and reduce, a filling level side-play amount FP is a negative value.

When the clock that extracts during than transmitting terminal professional slow, then filling level increases, fill a level side-play amount FP and be on the occasion of.

FP obtains a control signal V after digital-to-analogue conversion, V also is the function of time T.

This V can be regarded as the output of the phase discriminator of above-mentioned PLL mechanism.

V eliminates the randomized jitter composition of factor introducings such as network delay variation after low-pass filtering.

Because V is relative more weak signal, need obtain Vc with certain ratio of gains amplification.Vc makes voltage controlled oscillator produce the certain frequency increment as the input of voltage controlled oscillator, with the synthetic tranmitting data register that obtains E1 of the centre frequency (2.048Mhz) of oscillator.

Make Δ f ¹Be the frequency increment of this clock adjustment, then Δ f ¹It is the function of filling level side-play amount FP.

Substantially can represent with following formula:

Δf ¹＝4.096w*FP/0.5

The input of above-mentioned voltage controlled oscillator should be able to make the output of voltage controlled oscillator produce Δ f ¹Adjustment amount.

Through after the feedback regulation repeatedly, fill level basicly stable equal 0.5, half occupied state just, or be that the center is with minimum amplitude vibrations with half occupied state.At this moment, system clock is in the lock state, and local service clock and remote service clock mate substantially.

(2) based on RTP time stamp pattern

Say that intuitively this pattern is to utilize suitably mechanism equally, produces a frequency increment Δ f, the synthetic business clock f that obtains this locality of the centre frequency of this increment and voltage controlled oscillator _r

Figure 20 is based on the functional schematic of the adaptive clock recovery module of RTP time stamp pattern.

With fill based on buffering area the level pattern different be that two ends all need adding up the local service clock

With one 32 digit counter to the local service clock count.

When the TDM time slot is complete enter buffering area after, the value that writes down local business clock counter is in order to generating the time stamp of packet correspondence.

When transmitting terminal need send packet (is that unit sends with the time slot group), insert the local service clock count value of aforementioned record in Real-time Transport Protocol time stamp territory.

When receiving terminal when the Packet Based Network side joint is received a packet, extract time stab information from this packet, and this time stamp sent (Packet Based Network side) time stamp records the time stamp buffering area by sequence number relevant position as packet.

Represent this time stamp with Tts (n).

Write down the value of local local service clock counter simultaneously, as the reception time stamp of this packet.

Represent this time stamp with Rts (n).

The Huan Chong district is identical with going to tremble, and the time stamp buffering area also is a loop structure.T be can deposit and time stamp and T reception time stamp sent.Sequence number is that the time stamp of n is determined by n%T in the memory location of time stamp buffering area.

When local terminal received the 1st packet, opening entry sent time stab information.

When jitter-removing buffer by full sky during to half occupied state, E1 P12 read pointer points to the 1st byte of the 1st time slot of the 1st effective time slot group.This moment, the time stamp buffering area was deposited L/2 reception time stab information and L/2 transmission time stab information.

Make t (n) expression P12 read pointer point to the individual effectively moment of the 1st byte of n time slot of time slot group of n.

Make Δ Rts (n)=Rts (n+1)-Rts (n),

And Δ Tts (n)=Tts (n+1)-Tts (n).

Make Δ t (n)=Δ Tts (n)-Δ Rts (n),

Make Δ t ¹(n)=Δ t (n)/Δ Tts (n),

Δ t then ¹(n) phase discriminator that can be regarded as above-mentioned PLL mechanism is exported, and utilizes this output through loop filter, produces local business clock at the control voltage controlled oscillator.

Also so to say that according to this Δ t ¹(n) produce Δ f ¹And this Δ f (n), ¹(n) with synthetic t (n) the local service clock constantly that obtains of the centre frequency of local oscillator.

Δ f ¹(n) can determine by following formula with the relation of time stamp:

Δf ¹(n)＝4.096 w*Δt(n)/ΔTts(n)

The filtering logic can be used weighted moving average algorithm or exponential weighting moving algorithm.

13) the far-end type is determined

This step is to be used for determining whether the far-end of communication has FEC mechanism with it.

Because two reserved fields of control word are defined as FEC type field and FEC enable field respectively, so it is more convenient to judge whether far-end has identical FEC mechanism.

When the FEC enable field is 00, and the FEC type field is 0000 o'clock, can judge that far-end does not have corresponding FEC mechanism.Like that, local terminal also needs to forbid FEC mechanism, needs to do following processing:

The FEC enable field is made as 00 during package, and the FEC type field is made as 0000.

Length field is made as complete 0.

Do not generate the FEC bag.

Sequence number no longer is divided into two-stage.

During package, being not unit with the time slot group, is unit with the time slot.

Application example of the present invention

The multi-service grouping platform product M8000/T equipment of Wuhan Fenghuo Network Co., Ltd. development has been finished the function that on Ethernet high-efficiency reliable transmits E1.

Figure 21, be the positive facial contour of M8000/T equipment, 4 100 m ethernet light mouths are arranged, 4 100 m ethernet electricity mouthful and 4 gigabit Ethernet light mouths.These Ethernet interfaces can be used for remote equipment interconnected.

Figure 22 is the back side profile of M8000/T equipment, and 1-32 E1 interface arranged.

As shown in figure 23,3 M8000/T are barricaded as the network of linear topology.

Among Figure 23, the 2301st, the E1 interface connects 2M BER tester and M8000/T equipment.

The 2302nd, M8000/T equipment.

The 2203rd, the background traffic input interface, from the input of gigabit interface, maximum can be imported the flow of 2G.

The 2304th, the gigabit interface of M8000/T equipment.

2305 expressions couple together two equipment with the gigabit interface.

Fig. 9 is that system configuration of the present invention is described,

Be the control plane of this equipment on pci bus, comprise CPU, RAM memory, ROM memory, FLASH memory, out-of-band ethernet port PHY or the like.

Under pci bus is the datum plane of this equipment, comprises Ethernet switching chip, RAM memory, phy interface chip, FPGA, E1 transceiver or the like.

The E1 code stream that enters from the E1 transceiver is finished mapping to the Ethernet data bag at FPGA, comprises and realizes the FEC function.Enter the ethernet network from the Ethernet interface of appointment again through Ethernet switching chip, arrive the other end through network again.

At the other end, E1 packet and FEC bag enter exchange chip from Ethernet interface, arrive FPGA again, are reduced into the E1 code stream after FPGA handles, and send from the E1 transceiver again.

Figure 10 is that functional structure of the present invention is described.

Transmitting terminal has the transmission buffering area.

Receiving terminal has jitter-removing buffer and error correction buffering area.

Have the general data bag and the FEC of intermediate conveyor wrap.

Figure 11 is a key data stream description of the present invention.

Realize the processing sequence of a functional module in the arrow presentation function structure.

Dotted portion is represented the flow direction of E1 information in whole system.

During transmission, E1 information enters from the E1 interface of upper end, flows through after the functional module that needs, and is transformed into the Ethernet data bag, and the Ethernet interface from the lower end sends again.

During reception, the packet that carries E1 information enters from the Ethernet interface of lower end, and after the functional module that needs of flowing through, from Ethernet data bag reduction E1 information, the E1 interface from the upper end sends again.

Test Design as shown in figure 23.

Two gigabit ethernet ports input background traffic from the M8000/T equipment of centre, to produce to a certain degree congested, flow is imported from the Smartbits instrument, the size of may command background traffic.

Three equipment are interconnected with the gigabit Ethernet mouths, and middle equipment has used two gigabit Ethernet mouths, and the equipment on both sides then has used a gigabit Ethernet mouth.

E1 outputs to Error Detector from the Error Detector input and from the other end.

The two-way phone is connected to the E1 interface at two ends in the equipment by protocol converter.

At first the end to end performance that E1 transmits is forbidden under the situation of FEC mechanism in test again on Ethernet.

Test is again under the situation of using FEC mechanism then, the end to end performance that E1 transmits on Ethernet.

Test result is as shown in the table:

The background traffic input	1.0G	1.1G	1.2G	1.3G
					The E1 end to end performance of no FEC mechanism	A small amount of error code Error Detector AIS lamp flicker phone can put through but call voice is noisy	The long bright phone of a large amount of error code Error Detector AIS lamps can't be put through	The long bright phone of a large amount of error code Error Detector AIS lamps can't be put through	The long bright phone of a large amount of error code Error Detector AIS lamps can't be put through
E1 end to end performance with FEC mechanism	It is good that no error code Error Detector indication normal telephone can be put through call voice	It is good that no error code Error Detector indication normal telephone can be put through call voice	It is good that minute quantity error code Error Detector indication normal telephone can be put through call voice	It is good that minute quantity error code Error Detector indication normal telephone can be put through call voice

Sum up

E1 occupies quite great proportion in current network, it represents the application scale of existing tens dollars of technology SDH.

But existing network is trend of the times to packet network evolution.It also is inevitable and necessary that Packet Based Network transmits E1.

It is a contradiction that is difficult to be in harmonious proportion to the high request of QOS that Packet Based Network is propped up more weak and E1 to QOS.This contradiction has seriously hindered the development that Packet Based Network transmits E1.

The present invention is devoted to utilizing the FEC technology to improve the performance performance that Packet Based Network transmits end-to-end E1.Under the unwarrantable situation of network QOS, utilize the FEC function at two ends, the data-bag lost that network is caused carries out error correction and recovery.Thereby guarantee, even under the relatively poor situation of network performance, also can finish the correct transmission of E1 on Packet Based Network.

Simultaneously, the present invention guarantees and the similar technical compatibility that does not have FEC mechanism.In other words, have FEC mechanism interface can with the same class interface intercommunication interconnection that does not have FEC mechanism.

In enforcement of the present invention and application process, test shows, FEC mechanism of the present invention has good performance.

Although specifically show and described the present invention with reference to the preferred embodiments of the present invention, but persons skilled in the art should be understood that, under the situation in not breaking away from the principle and scope of the present invention that claims limit, can carry out concrete modification on form and the details to it.

Claims (32)

1. one kind is adopted forward error correction FEC mechanism to realize the method that reliable end-to-end E1 transmits on Packet Based Network, it is characterized in that the realization of this mechanism comprises following steps:

A. enter source node one end of Packet Based Network at E1, E1 is packaged into a data packet sequence, and according to FEC data packet group of this data packet sequence column-generation;

B. this sequence of data packet and FEC data packet group are sent to destination node one end that E1 leaves Packet Based Network by Packet Based Network;

C. in described destination node, if there is data-bag lost, then the packet of losing according to FEC package and sequence of data packet regeneration extracts or the regeneration E1 net load and the E1 code stream that reduces then.

2. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 1, it is characterized in that, can backward compatibility do not contain the similar technology of FEC mechanism, have FEC mechanism source node can with traditional destination node interconnection.

3. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 2, it is characterized in that, determine in the mode of auto-negotiation whether destination node possesses identical FEC mechanism, do not possess identical FEC mechanism as destination node, then source node is arranged to automatically non-FEC mode of operation.

4. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 3, it is characterized in that what transmit is the primary group of time division multiplexing tdm on Packet Based Network, speed is 2.048Mbps.

5. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 4, it is characterized in that the encapsulated layer of Packet Based Network has adopted real-time Real-time Transport Protocol, Real-time Transport Protocol is in order to carry the time stab information of mapped data bag.

6. realize the method for reliable end-to-end E1 transmission on the Packet Based Network according to claim 5, it is characterized in that the FORMID of the control word of the encapsulated layer of Packet Based Network adopts and expands definition, and in order to the machine-processed rule of using of sign FEC.

7. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 6, it is characterized in that the length field of the control word of the encapsulated layer of Packet Based Network is in order to the unidirectional end-to-end packet loss of sign far-end measuring.

8. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 7, it is characterized in that the sequence number of the control word of the encapsulated layer of Packet Based Network is defined as two-dimensional structure: chief series number and subsequence two-stage.

9. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 8, it is characterized in that the packet switching network PSN layer of Packet Based Network adopts IP agreement and frame format.

10. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 9, it is characterized in that the data link layer of Packet Based Network adopts ethernet link layer protocol and frame format.

11. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 10, it is characterized in that, design 32 business clock counters to the local service clock count, design 32 reference clock counters simultaneously outside reference clock is counted.

12. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 11, it is characterized in that, during the mapping (enum) data bag, the E1 code stream is pressed the time slot grouping, after a complete time slot entered buffering area, the value of record traffic clock counter and reference clock counter was to be used to generate the time stamp of this time slot.

13. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 12, it is characterized in that described A step also further comprises following substep:

During the mapping (enum) data bag, the RTP field need be carried definite time stab information in all packets that the mapping of time slot group generates, if be configured to the self-adaptation clock pattern, then time stab information generates according to the business clock count value of record; If be configured to the differential clocks reset mode, then time stab information is according to the business clock count value and the external reference clock count value of record, and generates according to certain rule.

14. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 13, it is characterized in that described A step also further comprises following substep:

One group of E1 time slot concentrated be mapped to packet data package, and data packet head carries the information such as timeslot number k that one group of E1 time slot comprises, wherein the number of the E1 time slot that comprises in the E1 time slot group can be configured to 8,16 or 32.

15. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 14, it is characterized in that described A step also further comprises following substep:

One group of E1 time slot is concentrated when being mapped to packet data package, can be disposed and use each time slot independently to be mapped to the independent mapping mode of bag, also can select each the time slot selected part data from one group of E1 time slot, be packaged into the synthetic mapping mode of packet again.

16. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 15, it is characterized in that described A step also further comprises following substep:

One group of E1 time slot is concentrated when being mapped to packet data package, when using synthetic mapping mode, selected corresponding position to be combined into byte or word from each byte of each time slot, recomposition frame, this mapping method can be used for application that information security is had relatively high expectations.

17. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 16, it is characterized in that described A step also further comprises following substep:

With one group of E1 time slot concentrate be mapped to packet data package after, generate the FEC package according to the grouped data packet sequence that generates.

18. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 17, it is characterized in that described A step also further comprises following substep:

Generate the FEC package, adopt Reed-Solomon (n, k) coding.

19. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 18, it is characterized in that described A step also further comprises following substep:

Generate the FEC package, adopt Reed-Solomon (n, k) during coding, k can be configured to 8,16 or 32, n can be arranged to k+4 by automatic negotiation, k+8, k+16, k+32 equivalence, and comprise the value of configuration information: n and k in the data packet head.

20. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 19, it is characterized in that described A step also further comprises following substep:

Generate the FEC package, the i combination of bytes that will comprise each time slot in the time slot group of k time slot obtains the k byte sequence together, and generates the error correcting code of n-k byte according to generator polynomial, respectively as the 1st, and 2...n-k the i byte that FEC wraps.

21. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 1, it is characterized in that described B step also further comprises following substep:

When sending packet, the transmitting time of general data bag depends on the time slot corresponding service clock count value that packet or its carry, that is to say that the difference of the transmission time interval of two adjacent data bags and two adjacent business clock count values is proportional in the time slot group.

22. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 21, it is characterized in that described B step also further comprises following substep:

When sending packet, in the transmission at interval of two general data bags, send the FEC packet, the FEC packet number that each time interval sends, value according to n and k determines that the time that sends the FEC packet determines that principle is: the time interval that the adjacent data bag is sent is even.

23. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 10, it is characterized in that described C step also further comprises following substep:

For the control time delay reception timer is set at receiving terminal, the cycle of overflowing equals k*125us.

24. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 23, it is characterized in that described C step also further comprises following substep:

The packet that receives removed behind the data packet head deposit jitter-removing buffer in according to the bag sequence number,

If be configured to self-adaptive recovery pattern based on time stamp, time stab information in the Real-time Transport Protocol that then needs packet is carried deposits the far-end time stamp memory space of buffering area in, utilize the local service clock to generate time stamp simultaneously, and this time stamp is deposited in the local time stamp memory space of buffering area;

Recover pattern if be configured to differential clocks, then obtain the local service clock in conjunction with external clock with this time stamp.

25. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 24, it is characterized in that described C step also further comprises following substep:

Deposit the error correction buffering area in according to the bag sequence number after the FEC that receives bag removed data packet head.

26. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 25, it is characterized in that described C step also further comprises following substep:

When the reception timer overflows, check the data of jitter-removing buffer, as have information dropout, the information of then utilizing error correction information regeneration to lose.

27. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 26, it is characterized in that described C step also further comprises following substep:

When the reception timer overflowed, drop-out then carry out clock recovery, and the method for clock recovery can be configured to adaptive mode or differential mode according to whether having same reference clock after recovering to finish.

28. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 27, it is characterized in that described C step also further comprises following substep:

When adopting the adaptive mode recovered clock, can be configured to based on the filling level of jitter-removing buffer or two kinds of patterns of time stamp of carrying based on Real-time Transport Protocol.

29. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 28, it is characterized in that described C step also further comprises following substep:

When adopting the adaptive mode recovered clock, business clock accuracy rating that can configuration restore.

30. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 29, it is characterized in that described C step also further comprises following substep:

If when being configured to the adaptive mode recovered clock, and be configured to use when cushioning the filling level, then recover the local service clock according to the operating position of buffering area;

If when being configured to the adaptive mode recovered clock, and when being configured to use time stamp, then the time stab information according to buffer stores recovers the local service clock.

31. realize the method that reliable end-to-end E1 transmits on the Packet Based Network according to claim 30, it is characterized in that described C step also further comprises following substep:

After clock recovery was finished, according to business clock, the data in binding buffer district were recovered the E1 information flow.

32. an application rights requires the equipment of each described method to realize that on Packet Based Network reliable end-to-end E1 transmits among the 1-31.

Images