Using CESoP to implement voice over an Ethernet passive optical network - Embedded.com

Using CESoP to implement voice over an Ethernet passive optical network

Ethernet passive optical network (EPON) technology has emerged as a keysolution for delivering triple-play services to residential andbusiness users over packetbased fiber access systems.

While data and video will benefit from the increased bandwidthafforded by EPON, the voice component is not as greatly served in the move topacket networks.

This article discusses the challenges in deploying voice-over- EPONand introduces Circuit Emulation Services-over-Packet (CESoP)as a potential solution.Time-sensitive voice traffic requires far less bandwidth but much morecare than data services.

Voice latencies are helped by plentiful bandwidth and QoS mechanisms in EPON, butvoice is sensitive to both short- and long-term network impairments.Network congestion, insufficient queue resources in an Ethernet switch,a routing change, or a more serious impairment such as a link or pathfailover will result in lost packets and degraded voice quality.

To help, voice-overpacket receivers are equipped with packet-lossconcealment schemes. Voice is also synchronous. A PBX must besynchronized to the PSTN to deliver T1/E1 service.

Otherwise, the framers in a PBX will incur periodic buffer slipsthat will affect QoS. Even for a plain old telephone service (POTS)line, if the receiver is not correctly synchronized to the sender,repeated, periodic buffer slips will appear as a burst of lost packets.

Enabling business services
Previous PON flavors were based on ATM technology , which included AAL1 support for circuitswitchedservices. The absence of this support within EPON means equipment isunable to provide a complete offering, including T1/E1 access, leasedline or private line services, for business users.

CESoP is an established technology that tunnels circuitswitchedtraffic across a packetswitched network. CESoP is a technology matchfor EPONs, allowing support for T1/E1 services for businessesas well as Nx64kbps residential voice service. CESoP packetizes the time division multiplex (TDM)traffic or voice samples, including the data and signaling, as eitherunstructured private line data or structured Nx64kbps voice channels.

These packets are then transported across a packet network, such asan EPON. At the far end, the ingress packets are smoothed using areceive jitter buffer. The TDM circuits or voice samples are thenextracted from the packets and played-out onto the TDM circuit.

T1/E1 service
An optical network terminal (ONT) for business users provides many10/100Mbps Ethernet interfaces, but does not directly support T1/E1services. Instead, a business office has many digital handsetsconnected to a PBX, which is connected through a T1/E1 line to thePSTN.

Figure1. By adding a CESoP block, the ONT has been upgraded to support T1/E1services, whether fractional, leased line or private line.

CESoP is a natural choice to service this market with EPON. Byadding a CESoP block, the ONT has been upgraded to support T1/E1services, whether fractional, leased line or private line (Figure 1, above ).

CESoP meets the challenges in providing voice and T1/E1 servicesover a packet-based network. Packetization latency may be as low as125┬Ás (one frame) or a larger 1ms (eight frames), especially in arelatively non-bandwidth-constrained network such as EPON.

The packet network latency itself across an EPON should be less thana few milliseconds upstream and even less downstream.

Figure2. An ONU targeting residential service may support voice service byconnecting a VoIP phone through its Ethernet interface.

The receive jitter buffer latency is matched with the EPON andwould be in the order of a couple milliseconds in the upstreamdirection and less in the downstream direction.

In total, one-way latency may be in the order of 5 ms upstream and3ms downstream. Measured lab values were 1,900 microseconds upstreamand 800 microseconds downstream, even under loaded data trafficconditions (with one frame packetization).

The timing and synchronization aspect of T1/E1 service has beenincorporated into the CESoP standards. Timestamps and sequence numbersare used in the CESoP packet headers to transfer timing informationfrom the PSTN to the customer premises equipment. In a plesiochronousdigital hierarchy environment, each T1/E1 connection may haveindependent timing via a different clock source.

Figure3: For an ONU with 32 analog POTS lines (32 x 64Kbps), the CESoPconnection bandwidth may be 2.5Mbps when taking the packet headeroverhead into consideration.

Each direction of an individual T1/E1 connection may also haveindependent timing. CESoP allows operation where each T1/E1 issynchronous with each other or where they are all asynchronous.

When voice packets are late or lost in the EPON, the CESoPconnection will replace those missing packets with appropriate “filler”data to minimize the effects on voice quality.

Best performance of CESoP connections across an EPON may be achievedin several ways. Selecting a low frames/packet value reduces overalllatency, and using a managed QoS-aware switch and enabling QoSmechanisms in the EPON prioritizes CESoP traffic over background datatraffic.

This may require the use of the virtual LAN protocol in the CESoPpacket header and/or the setting of the ToS bits in the IP header. TheCESoP inter-working function jitter buffer should also be sized to avalue suitable for the packet delay variation of the packet network -too small and there will be packet discard; too big and there will beunnecessary latency.

Figure4: VoIP typically implements three main blocks: voice processing,packet processing, and control and signaling.

CESoP for POTS
An optical network unit (ONU) targeting residential service may supportvoice service by connecting a VoIP phone through its Ethernetinterface, or using VoIP or CESoP through one or more POTS interfacesto provide traditional analog phone service.

Figure 1 earlier shows anONU implementation that uses CESoP to provide POTS. The choice of VoIPor CESoP depends on the application, complexity and cost.

CESoP is as a very simple mechanism to transparently carry a fewvoice channels from a residential customer to the PSTN, and providesseveral advantages versus VoIP. VoIP typically implements three mainblocks: voice processing, packet processing, and control and signaling (Figure 2, earlier ).

In comparison, CESoP removes the voice processing block entirely.Thus, it simplifies the complexity, and lowers the cost of the hardwareand software in the customer equipment.

CESoP tunnels the voice channels back to the PSTN-connected opticalline terminal, avoiding the need for local intelligence to handlecall-control processing and gateway signaling functions.

Thanks to the abundant bandwidth available in EPON, a CESoPconnection to carry all the voice channels provided by an ONU may beestablished permanently.

For an ONU with 32 analog POTS lines (32 x 64Kbps), the CESoPconnection bandwidth may be 2.5Mbps when taking the packet headeroverhead into consideration.

The use of CESoP over EPON enables voice services for residentialand business applications. CESoP may be used to provide fractional,leased line or private T1/E1 service for business customers or Nx64kbpsvoice channels over POTS lines for residential customers.

Focusing on business customers, the ability to provide T1/E1 serviceto a PBX is critical to offering a complete triple-play package. Thequality of the CESoP connections allows EPON to effectively deliver thesame voice performance and quality as a pure TDM network.

Peter Meyer is ApplicationsEngineer at Zarlink SemiconductorInc. For a PDF verlsion of this article go to Deliveringvoice services over EPON

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.