Challenge voip-Packet voice service based on circuit simulation

Compared with VoIP, packet network circuit simulation business has more flexibility, shorter delay and simpler features, is the most competitive alternative to VoIP technology. This paper expounds the characteristics of the packet network Circuit simulation service, and analyzes how to make the best use of these characteristics and the possible applications to operators and enterprises.

Despite the industry's concern over VoIP, the technology is not the only way to transfer voice through packet-switched networks, but it is not even the best approach in some applications. In addition to VoIP there is another option, that is, packet network circuit simulation Business (CESOP). Cesop was originally applied to an ATM network, it was necessary to apply the technology to the packet network with great changes. The ATM network circuit emulation business is now widely used, but the technology has not attracted enough attention in the industry, primarily because of the extra overhead of emulating a TDM circuit on an ATM network, and now the cost and efficiency advantages of using the circuit emulation business to transmit voice by packet are achieved.

Figure 1: The circuit emulation business implemented by packet switching network.

At present, Ethernet has been applied to metropolitan area Network, which has the characteristics of high bandwidth and low cost, and can integrate voice, video and data communication. However, Metropolitan Ethernet is unlikely to cover the area that TDM technology can extend to, after all, the application of TDM technology has a long history. In the communications industry, the "last Lane" is still the longest road. To become a complete service provider, a metropolitan Ethernet provider must extend its business to customers who are still outside their network, or to those who feel that there is no reason to abandon the high quality, reliable and economical TDM business.

The concept of Cesop

Cesop's basic idea is to build a "channel" on a packet-switched network in which TDM circuitry (such as T1 or E1) is implemented, so that the TDM device at either end of the network does not have to care whether the network it is connected to is a TDM network. Packet switched networks are used to simulate the behavior of TDM circuits and are therefore referred to as "circuit emulation" (Fig. 1).

The circuit simulation requires the interactive connection function at both ends of the packet switching network. At the entrance of the packet switching network, the interactive connection function converts the TDM data into a series of groupings, and the TDM circuit is regenerated by this series of packets at the exit of the packet switching network. There are two basic ways to implement this interactive function module, including structured emulation and unstructured emulation or structure agnostic (structure-agnostic).

The structured method uses the time slot structure inherent in the TDM circuit. First, the frame structure (f-bit in DS1) is extracted from the data stream, and then each time slot is added sequentially to the grouped payload, followed by the same time slot in the next frame, so repeated. When the payload is fully populated and a group header is added, the packet is sent to the packet switched network. Payloads typically contain approximately eight frames of TDM data (one DS1 contains 192 eight-bit groups). At the exit of the packet network, the TDM data stream is restarted and the new frame structure is used.

Unstructured transmission ignores any structure that may exist in the TDM circuit and considers the data as a pure bit stream for a given data rate. A series of eight-bit groups are intercepted sequentially from the TDM bit stream to form the payload of the packet. Therefore, the number of eight-bit groups that make up each packet payload is random. In general, the length of the payload is selected to make the grouping time around 1ms, for the T1 circuit, the length is 193 eight-bit group (see Figure 2).

Features to consider for packet networks

Figure 2: Structured circuit emulation service and unstructured circuit emulation service

Packet networks have features that are not the same as TDM networks, which can affect the performance of all circuit emulation systems. The basic features that a packet network must consider are packet loss, grouping reordering, and packet latency deviations.

Packet loss causes a short break in the TDM data stream, so a certain remedy must be taken, or the end-to-end latency of the emulation system will change. Normally, the Cesop interactive connection inserts data that is equal to the length of the missing data in the TDM stream of the output. The contents of this insertion data are random and vary for different applications. For example, if the emulation circuit transmits a voice service, the interpolation method is used to predict the content of the time slot. However, if the circuit transmits data, then the interpolation method is invalid, and a fixed value can be used at this time.

Grouping can also be reordered within a network. For example, if some groups pass a faster path through the network, they will reach the target device before the packets that are routed first. The Cesop interaction must be able to sort the groupings in the correct order before the TDM data stream can be regenerated.

Finally, even if all groups are routed through the same path of the network, there may still be some time deviations when they reach the interactive module at the exit of the network, which is "packet latency deviation" or "packet jitter". Because the TDM circuit has a constant bit rate, the fast arriving packet must be buffered before the output so that it can compensate for the delay between other slower groupings. This buffer is called the "jitter buffer".

Clock recovery

One of the most critical issues in any technology for circuit switching via grouping is clock recovery (see Figure 3). For example, when a dedicated leased line is used between two clients to connect through an emulation link on an operator packet network, the frequency fservice of the customer TDM business must be regenerated precisely at the exit of the packet network. A long time mismatch will cause the packet network exit to form a waiting queue, if the rebuilt clock is slower than the original clock, then the buffer is filled, otherwise it will be emptied. Both will result in data loss and quality of service degradation.

In a packet network, the packet is discontinuous in time, so the connection between its inlet and outlet frequency will be interrupted. Therefore, unless there is an external method to implement the clock allocation, the network interconnection function at the exit of the packet network must recover the original TDM service clock frequency by some means.

Figure 3: Simulation TDM service with clock recovery function

The clock recovery is the process of inferring the original clock frequency fservice from the changing packet arrival rate. However, as defined in ITU-T's g.823 and g.824 standards, the TDM network has very stringent requirements for clock stability. Therefore, the arrival rate needs to be filtered to remove the effect caused by packet delay changes. This approach is necessary because packet latency changes may contain extremely low frequencies of components. For example, packet networks are more expensive to use during the day than at night, resulting in daytime congestion and longer transmission delays. This causes a 24-hour fluctuation in the packet latency and may be fed to the frequency of the recovery clock.