Analysis and solution of the most popular factors

Analysis of factors affecting the performance of Enterprise IP and solutions

voip refers to a way of transmitting calls over IP networks. VoIP allows calls to be transmitted through existing IP data networks, thereby helping enterprises reduce communication costs. At the same time, the application of VoIP technology in the enterprise market makes data communication product manufacturers enter the voice market, and the fierce competition has spawned many new applications. Among many new applications, the integration of IP with enterprise data and desktop applications has become a hot spot. The voice quality of enterprise level IP is crucial to the quality of the whole fusion solution. At present, it is the fastest-growing variety in the foreign synthetic foam

factors affecting voice quality

voip enters the market as a new technology, competing with the traditional PSTN network. As an alternative technology, its voice quality should be consistent with or better than that of PSTN. However, because the IP network was not originally designed to transmit voice data, compared with the traditional system, some technical difficulties need to be overcome, and the designer must face the challenges brought by this. Common problems include network delay, jitter, packet loss and acoustic echo

1. Network delay

in the traditional PSTN network, the loop delay of intercontinental long-distance call or satellite long-distance call is about 500~600ms. At this time, when one party speaks, it will take a period of time for the other party to hear and react, and the calls between the two parties are disconnected. In the ITU standard, it is suggested that the loop delay of the system should be more than 300ms when the dynamometer part is not installed horizontally. In the IP system, the loop delay includes the packing time of voice message, the delay introduced by voice codec and the delay introduced by processing network jitter. The system designer must balance all factors to make the loop delay of the system as small as possible and make the call natural and smooth

2. When jitter

IP packets are transmitted on the network, the transmission route of each packet may be different, and the transmission and forwarding time on each node may also be different. In a VoIP call, the transmission time of each voice data message in the network is different. Some messages will arrive at the receiver later or earlier than the expected time of arrival. When playing back and decoding IP voice messages, the decoder plays back voice data at fixed intervals. Late voice packets will cause the decoder to have no data to solve for a period of time, and early data packets may be discarded. Therefore, the system needs to introduce message buffer to remove network jitter and smooth the impact caused by network jitter. However, the introduction of too deep buffer will increase the loop delay accordingly. It is necessary to select an appropriate buffer size to remove the network jitter and increase the loop delay at the same time

3. Packet loss

the ideal network situation is that all sent voice messages can be normally received by the receiving end. However, when IP packets are transmitted on the network, individual processing nodes of the network may have insufficient processing capacity or limited bandwidth, and some packets may be lost. Although there are some mechanisms on the IP network that can retransmit lost packets when they are lost, voice data with high real-time requirements cannot use these mechanisms

some voice coding and decoding methods have built-in packet loss compensation algorithm. When the packet loss rate is not high, interpolation method can be used to compensate, so that the listener does not feel the loss of voice information. However, in networks with high packet loss rate, other methods must be used to control the impact of packet loss on voice quality

4. Acoustic echo

there are usually two kinds of echo in the two four wire conversion system: line echo and acoustic echo

the line echo is caused by the impedance mismatch when the two or four lines are converted. In this case, the voice of the caller is reflected when it is transmitted to the two or four wire conversion at the far end, and the conversation with the other party is sent back to the caller's ear. If the loop delay is relatively small (<50ms), it is difficult for the caller to detect the existence of echo, but when the loop delay is relatively large, the caller can hear his own words. The loop delay of IP calls is generally large, so there must be an echo suppression unit at the node with two to four current conversion

acoustic echo is caused by the acoustic coupling between playing and listening on the side, and the sound played by the loudspeaker is fed back to the earpiece or microphone after one or more reflections in the space (as shown in Figure 1)

Figure 1: generation of acoustic echo

in the design of Enterprise IP, because there is no two or four wire conversion in IP, the influence of line echo does not need to be considered. The influence of acoustic echo, especially in hands-free mode, is a difficult and main problem that system designers need to consider

many solutions promoted in the market claim to be able to achieve full duplex handsfree. But in fact, most algorithms can only support half duplex handsfree calls. When full duplex works in hands-free mode, both sides of the call can hear each other's voice even if they speak at the same time. If it can only support half duplex handsfree calls, when both parties speak at the same time, one party's voice is suppressed and eliminated, and cannot be heard by the other party. Even when the local background noise is relatively large, the other party's voice is suppressed and no sound can be heard

network delay, jitter, packet loss and acoustic echo are all important factors that affect the quality of IP calls. System designers must properly solve these problems. If chip suppliers can provide mature solutions to solve these common problems, system designers can focus on designing differentiated new businesses and shorten the time for products to market

tnetv1050 IP solution

ti provides a comprehensive VoIP solution, covering IP, home customs, carrier customs, as well as VoDSL and vocable. Ti provides a variety of solutions for Enterprise IP requirements, among which the SOC scheme of tnetv1050/tnev1055 is based on TI's TMS320C55x DSP series programmable DSP and enhanced 32-bit MIPS core. It uses advanced system architecture to provide high performance and low power consumption while taking into account the scalability of the system. It also includes rich peripheral interfaces to meet the requirements of designing enterprise level systems

figure 2:tnetv1050/1055 IP processor

tnetv1050 chip includes a built-in Ethernet switch and dual PHY, which can provide IP services on the desktop and a PC connection at the same time. The USB interface of tnetv1050 enables IP to connect a variety of USB devices, such as PDAs. Designers can also expand many other functions through ti's vlynq interface, such as Wi Fi module, hardware encryption, etc. The chip integrates LCD controller, codec and keyboard interface, which reduces the system cost

ti's IP software suite includes DSP core algorithm and supporting CPU software package. The complete DSP software includes a variety of encoding and decoding algorithms, VaD, acoustic echo cancellation algorithm, message playback module, complete statistical debugging information function and voice quality monitoring module. For enterprise applications, Ti provides a complete acoustic echo cancellation module, supports full duplex hands-free function, supports local tripartite conference function and broadband encoding and decoding algorithm

cpu software kit, including ti's CPU software and third-party software modules. It includes API module that controls DSP, module that supports call control, encryption framework module, protocol stack interface, and third-party protocol stack module

ti software and third-party software together provide a complete software solution required by IP. The optimized chip solution and dsp/cpu software greatly reduce the system design time. Ti's software provides systematic solutions to common problems encountered in Enterprise IP design

1. Dealing with jitter and delay

for jitter and delay, TI's DSP software has been optimized accordingly, using a compact system structure to eliminate unnecessary algorithm delay. At the same time, the adaptive algorithm of de jittering buffer is adopted, which can calculate the real-time jitter of the network and adjust the depth of the buffer according to the size of the real-time jitter. When the network is in good condition, such as in the local area of an enterprise, the de jitter buffer algorithm can reduce the buffer depth to reduce the loop delay

2. Handling of lost messages

ti's software package includes active and passive packet loss handling methods. The active algorithm is to add redundant information (RFC 2198) or forward error correction information (RFC 2733) to the transmission message. In this way, if a message is found lost at the playback end, the lost message can be recovered according to the corresponding information. When the lost message cannot be recovered by active method, it can be compensated by passive method. According to the previous voice information, a segment of voice can be fitted for compensation

it can be seen from Figure 3 that when the network packet loss is relatively high, better voice quality can still be obtained by using active intervention

3. Acoustic echo processing

to solve the acoustic echo problem of IP, we need to start from two aspects

first, designers need to adopt a mold structure with good acoustic characteristics to reduce the amplitude of acoustic echo and the part of nonlinear echo. Designers can get help by consulting relevant design companies. Ti's proposal package contains acoustic design guidelines and puts forward suggestions on some key points in acoustic design. The key points in the design include: increasing the distance between the speaker and the microphone as much as possible; The microphone and horn should have acoustic isolation in the mold; The microphone should be wrapped with a soft sponge and then fixed to the phone to reduce the vibration transmitted from the mold; Shock absorbing pads are used for the electronic experimental machine with a maximum experimental force of less than 5kn when the phone contacts the desktop to reduce vibration; Choose speakers and microphones with good spectral characteristics

secondly, the acoustic echo processing module should be able to remove the suppressed acoustic echo well, and can correctly handle the double talk at both ends without damaging the continuity of background noise. The acoustic echo suppression module of Ti public data shows that it can quickly converge the prediction parameters and effectively detect the speech at both ends by using fast Fourier transform and frequency-domain algorithm. The multi-level bidirectional nonlinear algorithm is adopted, and the spectral characteristics are used to compensate the background noise when the nonlinear algorithm works

real time voice monitoring scheme Piqua

one of the difficulties of VoIP is the uncertainty of IP network, and the voice quality of VoIP system largely depends on the quality of network. Network administrators need a way to monitor the quality of VoIP in real time and actively intervene in network configuration and VoIP settings

piqua is a real-time voice monitoring scheme launched by Ti. It estimates users' subjective feelings about the voice quality of VoIP system according to complex algorithms and reports them in real time. At the same time, it also provides important network statistics. The upper software and network administrator can intervene the system configuration according to the real-time voice quality and statistical information. For example, when the loss rate of the network is relatively large for a certain period of time, resulting in the decline of voice quality, the system management software can configure different levels of redundancy or forward error correction according to the information of Piqua to ensure voice quality. At the same time, the network administrator can also understand the status of the whole network according to the records of Piqua, find the reason for the high packet loss rate, adjust the routing table or network bandwidth allocation, and actively maintain the quality of the whole voice network. (end)