Comparison between CESoP and VoIP, voice connection technologies based on packet networks

Circuit Emulation Service over Packet (CESoP) uses a different and simpler method than VoIP to transmit voice and circuit switching over packet networks, while also supporting a wider range of applications. This article compares CESoP and VoIP to provide a reference for Chinese engineers when making selections.

Voice services based on IP networks were once a novelty, and users had to endure lower voice quality in order to save a few yuan in call charges. Today, high-quality packet network voice services have become a reality.

From circuit switching to packet services

CESoP uses the same circuit emulation method as the widely deployed ATM to provide a transparent transmission mechanism for packet network transmission of circuit switching services. TDM services include data and signaling, which are grouped into unstructured leased line data or structured Nx64kbps voice channels, and then transmitted through Ethernet, IP or MPLS packet networks. At the far end, the input packet data is smoothly buffered using the receive jitter buffer, and then the TDM circuit is extracted from the packet and fed to the TDM interface.

In contrast, VoIP uses relatively old technology to transmit TDM voice channels over IP networks. Specifically, a gateway is used to terminate the public switched telephone network (PSTN) signaling and voice services. The gateway then establishes a single channel connection over the IP network for each circuit. At the far end, another gateway converts the voice packet data back to TDM.

The efficiency of the two solutions can be compared based on the software and hardware required to implement the VoIP or CESoP interworking function (IWF). A generic IWF unit includes three main functions:

Speech processing: includes echo cancellation, compression, audio detection and generation, as well as voice activity detection (VAD) and comfort noise generation (CNG) for silence detection and suppression.

Packet processing: includes conversion between TDM and packet data (packetization and depacketization), implementation of packet network protocol stack, provision of jitter buffer to compensate for packet delay variations and clock recovery.

Control and signaling: Includes telephony control and signaling functions such as channel associated signaling (CAS) and common channel signaling (CCS), as well as packet network call control such as H.323 and SIP.

VoIP solutions usually require all three functions, while CESoP solutions do not require voice processing functions, thus reducing complexity, and CESoP solutions also significantly reduce call control and signaling requirements. VoIP solutions terminate PSTN signaling at each IWF unit, and then complete equivalent call control between two IWF units, requiring a complete signaling stack to be implemented between two IWF units or gateways using control protocols such as H.323 or SIP.

In contrast, CESoP solutions usually do not terminate PSTN signaling, but rather "tunnel" CAS or CCS transparently through the packet switching network. PSTN call control is still completed by existing TDM equipment, and no additional complex signaling stack is required between two CESoP IWF units. A CESoP connection is invisible to PSTN signaling and does not affect PSTN signaling.

VoIP solutions usually use compression to minimize the required packet network bandwidth. Unlike VoIP, which compresses the voice traffic, the bandwidth savings in CESoP solutions are achieved by using a single packet connection to relay multiple TDM channels simultaneously. VoIP may also require VAD/CNG functions for silence detection and suppression, while CESoP does not require such bandwidth saving mechanisms.

VoIP solutions also require voice echo cancellation, but for CESoP, since the delay caused by its packetization process is small, the end-to-end delay can be kept below the 25ms standard specified by ITU-T G.131 call echo control, so echo cancellation is not necessarily required.

Ensure voice quality

The success of any circuit transmission solution based on packet networks depends primarily on its ability to provide voice services with the same quality and reliability as traditional telephone services. In addition to the usual quality issues caused by delay, compression, and echo, IP-based networks also introduce problems such as packet loss and jitter buffer slippage.

The biggest impact on quality is the delay caused by packet processing of voice channels. VoIP solutions usually support one voice channel per packet network connection, with packet sizes ranging from 10ms to 40ms; CESoP solutions support multiple voice channels per connection, with packet sizes ranging from only 125μs to 1ms.

According to the G.131 standard, if an echo canceller is not used, the one-way delay must be less than 25ms. Or according to the one-way transmission time requirements of ITU-T G.114, the one-way delay of high-quality voice services must be less than 150ms. It can be seen that the delay introduced by CESoP is much lower than that of the VoIP solution.

Another problem is that synchronization is not usually considered a critical parameter for VoIP time, which can cause "clicks" due to jitter buffer slips. In CESoP solutions, time synchronization is a key requirement because data services must also be supported. CESoP uses adaptive or differential clock recovery techniques to ensure that the TDM circuits remain synchronized, thereby avoiding buffer slips.

Supported packet network types

The VoIP solution is mainly implemented using IP as the Layer 3 protocol, while CESoP can be equally well applied to IP, MPLS or Ethernet networks, and provides high flexibility for voice, fax, MODEM and data services that pass through IP or MPLS networks at the same time.

VoIP solutions are best suited for environments with limited bandwidth, where compression techniques and VAD/CNG functions are required to minimize the required packet bandwidth. CESoP solutions use Nx64 trunks to reduce bandwidth requirements. A CESoP connection that provides G.711 Nx64 channel trunks can achieve the same or even better bandwidth savings than multiple G.729 8 kbps VoIP connections.

Since VoIP has very low requirements on network quality, VoIP solutions can be implemented on controlled or uncontrolled networks (even the Internet), but there is no guarantee in terms of voice quality. CESoP solutions need to be on a controlled packet network to ensure that the network performance level is sufficient for data or voice services on simulated TDM circuits.

Cost Factors

Considering the cost of developing an IWF and upgrading existing TDM equipment, the implementation cost of a CESoP solution is lower than that of VoIP. For CESoP IWF development, the lack of voice processing reduces hardware requirements and intellectual property licensing fees. Compared with VoIP solutions that use H.323 and SIP call control, CESoP's transparency of CAS and CCS signaling also reduces the requirements of the IWF software signaling stack.

For end customers who have already made significant network investments, this transparency also means that existing TDM equipment can continue to be used. CESoP provides the additional functionality needed to migrate TDM services to packet networks without having to replace or significantly upgrade existing TDM equipment.

Conclusion

Both CESoP and VoIP solutions use packet switching networks to carry voice and circuit switching services. CESoP is easier to implement and can support a wider range of applications through any controlled network. At the same time, its voice quality can ensure that end users are provided with services that are no different from traditional PSTN.

By Peter Meyer

Application Engineer

peter.meyer@zarlink.com

Zarlink Semiconductor