CN1400804A - Speech transmission method of Ethernet port - Google Patents

Abstract

This invention discloses an Ethernet internet phoneme transmission method inside group in which hardware devices at both sides of a terminal apply the custom signaling pattern to realize the call process at Ethernet side to realize functions similar to PBX and the conversion function between TDM signal and Ethernet frame to realize phoneme relay enabling transmission of data, phoneme service inside the group (entity) in a same network to save cost and investment.

Description

A kind of voice transmission method of Ethernet interface

Technical field

The present invention relates to the implementation method of a kind of voice transfer aspect in voice communication field, particularly a kind of voice transmission method based on 10 or 100 MBPS Ethernet interfaces.

Background technology

Present stage, voice transfer is one of the most important business that will support of enterprise network (group user).To the extensive use of grouping technology such as IP, ATM on voice transfer, embody speech business invariably from the online voice transfer of traditional circuit switching in real-life critical role.

Current, the group user speech business is mainly provided by following several networking modes, mode one: adopt subscriber exchange PBX to communicate with telephone network PSTN through the E1 trunk line; Mode two: adopt subscriber exchange PBX to communicate with telephone network PSTN through E1 trunk line cross-over connection ATM net; Mode three: adopt subscriber exchange PBX after the E1 trunk line is by IWF (IP phone front end processor) conversion, to communicate with IP network.This three classes networking mode all relates to subscriber exchange PBX, and the conversation of (interior lines) can be finished through subscriber exchange PBX between group user, if but dial outside line then need take E1 trunk line cross-over connection public network and finish.Wherein to be different from the place of mode one be to adopt Packet Based Network transferring voice (being respectively ATM voice and ip voice) for mode two and mode three, comparatively cheap on the conversation price, and this is more obvious mode three times.

But, open speech business in this three classes mode and just seem very uneconomical for the less group user of number of users.Particularly current user data service amount constantly increases, promoted the fast development of Packet Data Network, no matter be company, enterprise still is that public network operation commercial city is to having poured into huge enthusiasm in the online communication of grouped data, and in the enterprising lang sound transmission of Packet Data Network, except with its with respect to the comparatively cheap price of telephone network PSTN, more main is at the enterprising lang sound of Packet Data Network, the transmission of comprehensive multimedia service such as data and video is trend of the times, Ethernet is realized that speech business has realistic meaning as early being present in the network of business group internal if can fully utilize for group user.

Summary of the invention

The objective of the invention is to construct a kind of voice transmission method, utilize the group internal Ethernet to realize voice trunking, have that cost is low, networking characteristics easily based on 10 or 100 MBPS Ethernet interfaces.

The voice transparent transmission method that is based on the 10/100Mbps Ethernet interface that the present invention proposes, be applicable to the less situation of enterprises number of users (typical case is in 24 users), in order to reduce time-delay, the individual user's voice sampling of the N of calling terminal (N＜=24) is encapsulated in the transparent transmission that realizes striding Ethernet in the same ethernet frame.Implementation procedure comprises following step:

1), be stored in upstream digital voice signal buffering area, is the #1 buffering area from the audio digital signals of TDM side N*64Kbps;

2), from the #1 buffering area, take out (the N*8bit of unit, corresponding to the sampling of N user's voice) data encapsulation in same ethernet frame, deposit row buffer on the ethernet frame in, be the #2 buffering area, the nil pointer mode is adopted in this buffering area addressing, this buffering area omits to reduce the time-delay that buffering is brought, and is transferred to the opposite end through Ethernet;

3), be stored in the downlink Ethernet frame buffer zone, be the #3 buffering area from the ethernet frame of Ethernet;

4), take out ethernet frame from the #3 buffering area, parse audio digital signals and be stored in descending audio digital signals buffering area, be the #4 buffering area, the TDM signal that reverts to N*64Kbps is sent to the user.

Outstanding effect of the present invention is to make full use of the Ethernet that extensively exists at present to carry out voice transfer, realizes data, the transmission of speech business in consolidated network that group (enterprise) is inner, to save cost and investment.

Description of drawings

Fig. 1, Fig. 2, Fig. 3 are three kinds of mode schematic diagrames of background technology

Fig. 4 is the embodiment of the present invention schematic diagram

Fig. 5 is a flow chart of the present invention

Embodiment

Consult Fig. 4, if user N will converse with user M, the number of user N off-hook and appropriation family M so, hardware device 1 detect user N off-hook with and the number dialled after, with the message conversion of correspondence is the message format of Ethernet side, arrive hardware device 2 through Ethernet, the caller of announcement and called whom is, if called subscriber M free time then hardware device 2 control called subscriber M rings, otherwise the busy message of hardware device 2 loopbacks one called subscriber M is given hardware device 1, and hardware device 1 send busy tone to notify its on-hook for calling subscriber N.

The present invention is applicable to that the inner Ethernet that adopts of group user carries out voice transfer, and the equipment that adopt at the end points two ends adopts self-defining signaling format to realize calling procedure in the Ethernet side.In addition, hardware device 1 and hardware device 2 should be realized two-part function: can realize being similar to the translation function between subscriber exchange PBX function and realization TDM signal and the ethernet frame.

Consult Fig. 5, adopt Ethernet to realize that the transparent transport process of voice is described in detail as follows:

A, be stored in upstream digital voice signal buffering area, be the #1 buffering area from the audio digital signals of TDM (altogether N user, N＜=24) side N*64Kbps; The #1 buffer size is unit (i.e. the length of a frame) with N*8bit, a corresponding N user's 8bit sampling respectively;

B, from the #1 buffering area, take out a unit (N*8bit) data encapsulation in ethernet frame, deposit row buffer on the ethernet frame in, for the #2 buffering area, be transferred to the opposite end through Ethernet; The size of #2 buffering area is that (in fact the size of ethernet frame is between 64*8bit ~ 1518*8bit in unit with the size of 64*8bit ethernet frame, but because number of users N＜=24 that the present invention is suitable for, so adopt minimum ethernet frame size to get final product), i byte of frame data is corresponding to i the user (i=1 of TDM side, 2,3 ..., N).It is to reduce the time-delay that encapsulation process is brought that N user's voice sampling is encapsulated in the benefit of transmitting in the same ethernet frame.In addition, when if the TDM signal is encapsulated in the ethernet frame, the Ethernet side data sends and adopts the nil pointer addressing system (is after the #1 buffering area adds the data of full N*8bit, original position with this buffering area of transmission pointed of Ethernet side), the #2 buffering area can save the time-delay that brings because of buffering to reduce so.

C, be stored in the downlink Ethernet frame buffer zone, be the #3 buffering area from the ethernet frame of Ethernet; The size of #3 buffering area is that (in fact the size of ethernet frame is between 64*8bit ~ 1518*8bit in unit with the size of 64*8bit ethernet frame, but because number of users N＜=24 that the present invention is suitable for, so adopt minimum ethernet frame size to get final product), i byte of frame data is corresponding to i the user (i=1 of TDM side, 2,3 ..., N).

D, take out ethernet frame from the #3 buffering area, parse audio digital signals and be stored in descending audio digital signals buffering area, be the #4 buffering area, the TDM signal that reverts to N*64Kbps is sent to the user.The #4 buffer size is unit (i.e. the length of a frame) with N*8bit, a corresponding N user's 8bit sampling respectively, corresponding relation between ethernet frame data byte (as i byte) and TDM time slot (as j time slot) is calling terminal notice called end during by call setup, to realize the exchange process of time slot.For considerable index of voice transfer is time-delay, and the conference echogenicity of delaying time can influence voice quality when serious.This lagger should be controlled within 10 ~ 30mS (representative value is 16mS) during traditional commerce was used, the measure that should increase echo control if this value surpasses 50mS and eliminate.Calculate the index of delaying time with regard to processing procedure of the present invention below, this time-delay comprises the following aspects:

1) buffer delay

The TDM signal respectively cushions 1 time at calling terminal and called end transmitting-receiving, accumulative total 4 times, and each buffering causes the delay of a frame, i.e. 125uS (8kHz sampling); Ethernet frame respectively cushions 1 time at calling terminal and called end transmitting-receiving, accumulative total 4 times is (if when the TDM signal is encapsulated in the ethernet frame, the Ethernet side data sends and adopts the nil pointer addressing system, it then is 3 times), each buffering causes the delay of a frame, if the transmission rate of Ethernet is 10Mbps, then the time-delay of a frame is 64*8/10uS, so the accumulative total time-delay is:

125uS*4+64*8/10uS*4＝0.7048mS

In fact, if when the TDM signal is encapsulated in the ethernet frame, the Ethernet side data sends and adopts the nil pointer addressing system, and the transmission rate of Ethernet is higher than 10Mbps, and the buffering time-delay also will be lower than this value.

2) processing delay

This time-delay mainly is meant the time that the TDM signal is encapsulated into ethernet frame and parses the TDM signal from ethernet frame.Processing method has two kinds: 1, by CPU, software and hardware combining is transmitted; 2, fpga logic is realized the devices at full hardware forwarding.Wherein the delay of the 2nd kind of method is less than the 1st kind, and existing is that example is calculated processing time-delay with the 1st kind of processing method.If the work clock of CPU is 50MHz, data-bus width is 32bit, then CPU reads or writes frame data (TDM side one frame is 24*8bit from each buffering area, Ethernet side one frame is 64*8bit) the required time is (24 or 64)/4*20nS, calling terminal and called end TDM signal and ethernet frame respectively cushion 2 times, so handle maximum delay are:

24/4*20nS*4+64/4*20nS*4＝1.76uS

3) network delay

Because of the present invention is applicable to voice transfer on the business group internal Ethernet, so the forwarding point of frame transmission process is approximately 1 ~ 2, the time is approximately several milliseconds.

Take a broad view of three kinds of above time-delays, total time-delay in 16 milliseconds, need not the echo braking measure certainly.

Claims (2)

1, a kind of voice transmission method of Ethernet interface is characterized in that it may further comprise the steps:

1), be stored in upstream digital voice signal buffering area, is the #1 buffering area from the audio digital signals of TDM side N*64Kbps;

2), the data encapsulation of from the #1 buffering area, taking out a unit in same ethernet frame, deposit row buffer on the ethernet frame in, be the #2 buffering area;

3), be stored in the downlink Ethernet frame buffer zone, be the #3 buffering area from the ethernet frame of Ethernet;

4), take out ethernet frame from the #3 buffering area, parse audio digital signals and be stored in descending audio digital signals buffering area, be the #4 buffering area, the TDM signal that reverts to N*64Kbps is sent to the user.

2, method according to claim 1 is characterized in that described #2 buffering area addressing employing nil pointer mode, and this buffering area omits to reduce the time-delay that buffering is brought, and is transferred to the opposite end through Ethernet.

Images