VolP terminal and gateway

H.323 The protocol defines various VoIP equipment , It mainly includes terminals and gateways , They can be collectively referred to as terminal nodes . Such devices perform similar functions ： Send and receive to a call Mark IP port . In some cases , That is, multimedia configured with corresponding software PC Calling each other , It can be regarded as two H.323 Communication between terminals . In another case , That is, call between phones , Then we have to go through PSTN To IP The network goes to PSTN Transfer of . In both cases, it is necessary to describe the end node IP The inflow and outflow of voice data ,H.323 In the agreement, it is called information flow .

VoIP The main body of information flow is voice information ,H.323 The protocol allows various coding schemes, but in order to take care of compatibility , all H.323 Terminals and gateways must have ITU-TG.711 codecs , namely 64kbit/sPCM Editing and decoding machine , Although not efficient , but PCM At least the compatibility of different devices is guaranteed . in addition ,H.323 The protocol also defines some other speech coding schemes ：
（1）G.728
ITU-T Defined 16kbiu/s coding standard , Algebraic coding excitation with low delay - Linear prediction （LD-ACELP） Data compression algorithm .
（2）G.729 and G.729a
ITU-T Made up 8kbit/s coding scheme , Linear prediction is excited by a common bottle structure algebraic coding （CS-ACELP） Data compression algorithm .G.729a Than G.729 The algorithm complexity is small , Short execution time , Therefore, the algorithm has been favored by most gateway manufacturers .
（3）G.723.1
The standard defines 5.3kbit/s and 6.4kbit/s Two rates . The corresponding team leader is 20 Byte and 24 byte , Every time 30ms One shift rate switching can be selected . When the originator is idle ,G.723.1 Specify sending 4 Bytes of free data packets . And G.729 comparison ,G.723.1 It can better handle DTMF （DTMF） Signaling , That is, it makes better use of in band voice signaling .
These coding techniques are different due to the different mathematical models , So the effect of coding is also different . People use three technical indicators to measure whether these coding techniques are suitable for VoIP System ： Output data rate 、 Algorithm execution delay 、 Intelligibility of encoded speech signal .
Studies have shown that , The speech quality is inversely proportional to the compression rate . for example , With 64kbit/sPCM The speech quality is the standard 、 Then switch to other low rate algorithms , The speech quality of subjective evaluation has declined .
At first , Quantitative indicators of speech quality are difficult to test , Now? , People mainly use two kinds of evaluation indicators , Delay and average subjective score （MOS）. Data rate 、 Time delay 、 And MOS The relationship is as shown in the figure 7-6 Shown .

chart 7-6 Data rate delay MOS Diagram for

generally speaking , The higher the complexity of the algorithm , The higher the compression efficiency , But the longer the delay . Although gateway and terminal manufacturers try to use ASIC chips （ASIC） Or field programmable devices to shorten the processing delay , But it doesn't change this relationship .
Too long delay is extremely unfavorable . for example , Before the popularization of submarine optical cables , Transoceanic calls transferred via synchronous satellites often cause intolerable echo interference due to large time delay . Although people who have made such calls think 0.5~1s The delay of will make the call difficult , But in fact, most people are more than 0.2s The time delay is unbearable .
If the delay from sending to receiving is limited to algorithm processing , The situation is not too bad . But actually , Network processing node （ Router 、 Switches etc. ） Greater delay will be introduced . For the above reasons , The codec designer is forced to send voice of the highest quality with the shortest possible data stream , And as soon as possible .
Time delay index is important , But the average subjective score （MOS） Indicators may be more important .MOS It is the subjective criterion of whether speech is distorted , Given by a group of listeners . Perhaps the most attractive thing about speech coders is that they can 8% The bandwidth provided 80% The fidelity of .
In fact, some manufacturers are not willing to adopt the above standards . They boast 11 Say your vocoder is better than G.7xx Better . But mainstream manufacturers hold a negative attitude towards this , Because this will lead to incompatibility of the device .