Traditional mobile networks, such as today’s 2G (GSM, CDMA 1x) and 3G (UMTS/HSPA and EVDO) networks, are based on Time Division Multiplexing (TDM) for transmission. These TDM networks comprise the majority of the backhaul networks for transport of voice and data traffic. Figure 7-1 shows backhaul penetration worldwide by technology.1
Figure 7-1 Backhaul Network Penetration
As mobile network data traffic grows, and user demand and dependency on the mobile operator as a data access provider increases, mobile operators are exploring various offload mechanisms to migrate legacy TDM networks to modern Ethernet and IP. Figure 7-2 demonstrates the increased bandwidth requirements per base station, to support next-generation mobile services.2
Figure 7-2 Backhaul Bandwidth Requirements
This migration allows a mobile operator to shed excess Operating Expenditures (OPEX) associated with TDM transport. However, during this migration, supporting legacy TDM interfaces and network elements is critical for continuing operations.
Various IP-based offload mechanisms may be employed to allow for this migration without the high Capital Expenditure (CAPEX) outlay for new equipment (BTS, BSC, and MSC infrastructure).
These IP-based offload mechanisms can be largely categorized as follows:
- Backhaul offload involves encapsulation of standard TDM protocol communications between the Base Transceiver Station (BTS) and the Base Station Controller (BSC), the BSC and the Mobile Switching Center (MSC), or inter-BSC/MSC, into IP packets.
- Signaling protocol offload involves protocol conversion of signaling packets. An example of signaling protocol offload is SS7/SIGTRAN.
- Bearer protocol offload involves protocol conversion of bearer packets. Examples of bearer protocol offload include Transcoder-Free Operations (TrFO) mechanisms and IP Soft-Handoff mechanisms.
Backhaul Offload with Pseudowires
Pseudowires allow for the emulation of point-to-point or point-to-multipoint links over a Packet-Switched Network (PSN). Pseudowire technology provides a migration path, allowing an operator to deploy packet-switched networks without immediately replacing legacy end-user equipment.
Each pseudowire presents a single, unshared “circuit” for carrying “native” services, such as ATM, SONET/SDH, TDM, Ethernet, or Frame Relay, over the PSN. The PSN may either be Layer 2 Tunneling Protocol Version 3 (L2TPv3), MPLS, or generic IP.
Many standards organizations, including the Internet Engineering Task Force (IETF), the Metro Ethernet Forum (MEF), and the International Telecommunications Union Telecommunications Standards Sector (ITU-T), have defined the encapsulation techniques for transport of the relevant protocols in mobile networks today, as follows:
- IEEE RFC3985: Pseudowire Emulation Edge-to-Edge (PWE3).
- IEEE RFC5087 and ITU-T Y.1453: Time Division Multiplexing over IP (TDMoIP).
- IEEE RFC4553: Structure-Agnostic Time Division Multiplexing over IP.
- IEEE RFC5086: Circuit Emulation Services over Packet-Switched Networks (CESoPSN).
- IEEE RFC4717 and ITU-T Y.1411: ATM Pseudowires.
- IEEE RFC4842: Synchronous Optical Network/Synchronous Digital Hierarchy (SONET/SDH) Circuit Emulation over Packet (CEP).
Prior to discussing pseudowire technology itself, the following examples should help to clarify various uses for pseudowire technology in mobile networks. The examples discussed may not be applicable to all mobile operators or all mobile infrastructure vendors, but are representative of some of the many deployment scenarios where pseudowires have been successfully deployed as an offload mechanism. The examples cover four scenarios, as follows:
- TDMoIP Pseudowire for CDMA/EVDO or GSM Backhaul Networks
- CESoPSN Pseudowire for Inter-BSC/MSC Connectivity
- ATM Pseudowires for UMTS R4 Connectivity
- Pseudowires for Multi-RAN Environments
Details of each pseudowire technology and implementation follow.
TDMoIP Pseudowires for EVDO or GSM Backhaul Networks
As discussed in Chapter 4, “An IP Refresher,” the traditional mobile backhaul network for a CDMA or GSM network consists of TDM interfaces on both the Base Transceiver Station (BTS) and the Base Station Controller (BSC). These TDM interfaces connect to a backhaul provider’s T1/E1 circuits for transport. Figure 7-3 illustrates a mobile backhaul network with standard TDM backhaul.
Figure 7-3 Traditional TDM Mobile Backhaul Network
TDM pseudowire technology plays a key role in allowing mobile operators to migrate their backhaul networks between the BTS, or cell site, and BSC or MSC location. The pseudowire provides a “transparent wire” between these locations and preserves the integrity of the TDM framing as it is transmitted across the PSN. Figure 7-4 illustrates a mobile backhaul network that uses TDMoIP pseudowires for transport.
Figure 7-4 Mobile Backhaul Network with TDM Pseudowires
CESoPSN Pseudowires for Inter-MSC/BSC Connectivity
As discussed in Chapter 4, the traditional MSC and BSC functionality and connectivity is typically TDM-based. Interconnectivity between all BSCs/MSCs is essential for handling mobility of a voice session in a circuit-switched voice (GSM, CDMA 1x) environment. However, in order to support such an environment, typical mobile deployments rely on a combination of point-to-point TDM circuits between BSC and MSC, and fully-meshed or star configurations of TDM circuits from the MSC toward the core network. Figure 7-5 illustrates one such topology.
Figure 7-5 Inter-MSC/BSC Connectivity
The overall cost of maintaining a fully-meshed, point-to-point TDM architecture is significant from an OPEX perspective. By reducing the number of TDM circuits required from the Local Exchange Carrier (LEC), a mobile operator may immediately see impact to operating margins. One such way to reduce the number of circuits is to leverage CESoPSN pseudowires for interconnecting MSC and BSCs, as illustrated in Figure 7-6.
Figure 7-6 Inter-MSC/BSC Connectivity with CESoPSN Pseudowires
Inter-MSC/BSC connectivity with CESoPSN pseudowires allows a mobile operator to use existing infrastructure, namely their IP core network, for transport of voice traffic.
ATM Pseudowires for UMTS R4 Backhaul Networks
UMTS Release 4 networks rely heavily on ATM as a transport mechanism for data traffic. Similar to the model previously discussed for transport of TDM backhaul traffic in CDMA and GSM environments, fixed circuits must be deployed to allow for mobility. These fixed ATM circuits, known as Permanent Virtual Circuits (PVCs), are discussed in more detail in Chapter 4. Figure 7-7 depicts a UMTS R4 backhaul network, from Node B to RNC and from RNC to MSC/SGSN.
Figure 7-7 UMTS R4 Backhaul with ATM
By migrating to ATM pseudowires and leveraging IP core assets, mobile operators can simplify their architecture, reduce costs, and begin preparing for fourth-generation mobile technology deployment, such as 3GPP Long-Term Evolution (LTE), discussed in Chapter 3, “All-IP Access Systems.” Figure 7-8 illustrates one potential solution with ATM pseudowires.
Figure 7-8 UMTS R4 Backhaul with ATM Pseudowire
It is also possible that IP backhaul and IP core networks may converge over a common IP/MPLS network.
Converging Multiple RAN Technologies over Common Pseudowire
As mobile operators complete their transition from solely circuit-switched voice networks to voice and data networks, mobile networks begin to become an overlay of multiple radio technologies. With all these multiple overlays requiring unique circuits (TDM or ATM), mobile operators incur large OPEX for maintaining multiple different backhaul networks. For instance, a CDMA operator maintains a CDMA 1x voice network and EVDO data network simultaneously. Even if the radio access cards reside in the same physical element, mobile operators use unique circuits for voice and data traffic in order to facilitate troubleshooting and problem isolation.
Pseudowires present an opportunity for mobile operators to deploy a unified backhaul architecture while still managing each circuit individually.
Example 1, illustrated in Figure 7-9, highlights a converged RAN architecture for a CDMA operator.
Figure 7-9 Converged RAN Architecture for CDMA
Example 2, illustrated in Figure 7-10, highlights a converged RAN architecture for a GSM/UMTS operator.
Figure 7-10 Converged RAN Architecture for UMTS
Pseudowire Emulation Edge-to-Edge (PWE3)
Pseudowire Emulation Edge-to-Edge RFC 3985 provides the structure and architecture for emulation of Frame Relay, ATM, Ethernet, TDM, and SONET over packet-switched networks using IP or MPLS.
Pseudowires for Time Division Multiplexing (TDM)
At the most basic level, TDMoIP pseudowires segment T1/E1 frames, encapsulate these frames in Ethernet, and fragment the frames into IP packets for transport across the PSN. At the destination, the IP header is stripped, the Ethernet frame is decapsulated, and the original bit stream is reconstructed, including regeneration of clock information. Figure 7-11 illustrates a high-level view of a TDMoIP pseudowire.
Figure 7-11 High-Level View of TDMoIP Pseudowire
Structure-Awareness of TDM Pseudowires
TDM over IP pseudowires can be categorized into two classes, as follows:
- Structure-Agnostic Transport over Packet (SAToP): With structure-agnostic transport, the protocol may disregard all structures imposed on TDM signaling or framing. Therefore, this transport method is simply bit-by-bit transport. Structure-agnostic TDM over IP is standardized in RFC4553. The PE devices in SAToP transport network do not participate in TDM signaling and do not interpret the TDM data. This implies that there are no assurances that network degradation does not impact the TDM structure.
- Structure-Aware Transport over Packet: With structure-aware transport, such as TDMoIP and CESoPSN, the integrity of the TDM structure is ensured, even in cases of network degradation. Because PE devices have exposure to the TDM signaling, individual channels are exposed, allowing the network to utilize Packet Loss Concealment (PLC) and bandwidth conservation mechanisms on a per-channel basis.
A frame structure refers to the way a single communications channel is multiplexed in several individual channels. By multiplexing the underlying channel, more than one data stream may be simultaneously transmitted at a time. Because TDM is based on the time domain, a single frame is actually a constant-length time interval. Within this time interval, fixed-length timeslots, each representing a single circuit-switched channel, are transmitted.
A multiplexer is responsible for assigning data, or bytes, from a bitstream to each timeslot, and a demultiplexer is responsible for re-assembling the bitstream. Although every timeslot may not be used, the entire frame is always transmitted in order to ensure that frames remain synchronized.
A T1 frame consists of (24) 8-bit (1-byte) timeslots plus a synchronization bit, allowing for 193 bits. An E1 frame consists of 32 timeslots, each containing 8 bits, or a total of 256 bits per frame, including a synchronization bit. In both cases, frames are transmitted 8,000 times per second. With this framing information, it is easy to calculate the total available bandwidth for both T1 and E1 circuits:
- T1 Circuit Bandwidth = (24 timeslots * 8 bits + 1 synch bit) * 8,000 frames per second / 1*10^6 bits/Megabit = 1.544 Megabits per second
- E1 Circuit Bandwidth = 32 timeslots * 8 bits * 8,000 frames per second / 1*10^6 bits/Megabit = 2.048 Megabits per second
Multiple channels, each containing 8000 8-bit samples per second, are multiplexed together using TDM framing, as illustrated in Figure 7-12.
Figure 7-12 TDM Frame Multiplexing
Structure-aware emulation assumes that the TDM structure itself, including the framing and control information, are available to the pseudowire edge device. With this information available, pseudowire encapsulation can be done in a more intelligent manner, with the edge device selecting specific channel samples from the TDM bitstream. Structure-aware transport may ensure the integrity of the original TDM structure via three distinct adaptation algorithms, as follows:
- Structure-Locking: Structure-locking ensures that each packet on the pseudowire contains an entire TDM structure, or multiple/fragments of TDM structures. The exact number of frames included is locked for all packets, in both directions. The order of the frames in the PSN is the same as those within the TDM frame sequence. When a TDM bitstream arrives, consecutive bits from the bitstream, most significant first, fill each payload octet. Structure-locking is not used in TDMoIP.
- Structure-Indication: The structure-indication method is derived from ATM Adaptation Layer 1 (AAL1), described in Chapter 4. Unlike structure locking, structure indication allows for pseudowire packets to contain arbitrary-length fragments of the underlying TDM frames. These fragments are taken from the bitstream in-sequence, from the most-significant bit first. The pseudowire packets also include pointers to indicate where a new structure begins. Because the bitstream sequence is identical to the sequence contained in the PSN, this method is commonly known as “circuit emulation.”
- Structure-Reassembly: The structure-reassembly method allows for specific components of the TDM structure to be extracted and reorganized within the pseudowire packet structure by the ingress pseudowire edge, with enough information such that the other edge of the pseudowire may reassemble the original TDM structure. The structure-reassembly method allows for bandwidth conservation by only transporting frames/timeslots that are active. This method is commonly known as “loop emulation.”
TDMoIP uses the structure-indication algorithm for constant-rate, real-time traffic and the structure-reassembly algorithm for variable-rate, real-time traffic. CESoPSN uses the structure-locking algorithm.
Packet Loss Concealment (PLC)
TDM networks are inherently lossless. Because TDM data is always delivered over a dedicated channel at a constant bitrate, TDM bitstreams may arrive with bit errors, but are never out of order and never get lost in transit.
The behavior of a TDM network is not replicable in a cost-efficient manner over an IP network. Implementation of Quality of Service (QoS) and traffic-engineering mechanisms may be used to reduce traffic loss, but there is no guarantee that packets will not arrive out of order, or arrive at all. Packet-Switched Networks are inherently unreliable, and leverage higher-layer protocols to provide for sequencing, retransmission, and reliability.
Because TDM pseudowires carry real-time bitstreams, it is not possible to rely on retransmission mechanisms. Packet Loss Concealment (PLC) masks the impacts of these out-of-order or lost packets. In the case of lost packets, arbitrary packets are inserted into the bitstream to ensure that the timing is preserved. Because a TDM pseudowire packet is considered lost when the next packet arrives, out-of-order packets are not tolerated. TDM pseudowires use different types of arbitrary packets to conceal packet loss, as follows:
- Zero Insertion: Insertion of a constant value, or zero, in place of any lost packets. For voice, this may result in some choppiness.
- Previous Insertion: Insertion of the previous frame value in place of any lost packets. This method tends to be more beneficial for voice traffic, because voice tends to have a stationarity aspect. This stationarity means that the missing frame should have characteristics similar to the previous frame.
- Interpolation: Because a TDM pseudowire is considered lost when the next packet in sequence arrives, the receiver has both the previous and next packets upon which to base the missing frame value. Interpolation algorithms ranging from linear (straight-line interpolation of missing frame value) to more predictive (statistical calculations of missing frame value) may be used; however, there is no standard method for TDM pseudowire frame interpolation.
Time Division Multiplexing over IP (TDMoIP)
TDM over IP was first developed by RAD Data Communications in 1998, and first deployed in 1999 by Utfors, a Swedish broadband communications operator later acquired by Telenor.
The basic structure of a TDMoIP packet is depicted in Figure 7-13.
Figure 7-13 TDMoIP Packet Structure
TDMoIP packets are composed of three main parts, as follows:
PSN Headers: PSN headers contain IP, MPLS, L2TPv3, or Ethernet information required to send the packet from the pseudowire ingress device toward the destination device, or pseudowire egress device. For example:
- IP transport requires that the source/destination IP address and port number be included in the header.
- MPLS transport requires that the MPLS tunnel label be included in the header.
- L2TPv3 transport requires that the L2TPv3 Session Identifier (pseudowire label) be included in the header.
- Ethernet transport requires that the Ethernet source/destination MAC address, VLAN header, and Ethertype be included in the header.
- Control Word: The Control Word is included in every TDMoIP packet. The Control Word includes information on TDM physical layer failures/defects (local or remote), length of the packet (to indicate if the packet is padded to meet PSN minimum transmission unit size), and sequence number (for detection of lost or misordered packets).
- Adapted Payload: The pseudowire ingress device uses either structure-indication or structure-reassembly in order to fill the packet payload.
Defects in a TDMoIP network may occur in multiple different locations. Depending on the location of the defect, standard TDM OAM mechanisms or TDMoIP mechanisms may be used to alert the TDM peer. Figure 7-14 illustrates the multiple defect locations in a TDMoIP network.
Figure 7-14 TDMoIP Network Defect Locations
Table 7-1 includes information about the reference points illustrated in Figure 7-14 and correlated OAM mechanisms, if available.
Table 7-1. Reference Points and OAM Mechanisms
Defect in the L2 TDM network that impacts any number of circuits terminating on the pseudowire edge devices.
The defect is communicated to the pseudowire edge devices and the remote TDM peer via native TDM OAM mechanisms.
Defect on the pseudowire edge TDM interface.
Defect on the pseudowire edge PSN interface.
Defect on the PSN that impacts any number of pseudowires terminating on the pseudowire edge devices.
The defect is communicated to the pseudowire edge devices via PSN or pseudowire OAM mechanisms.
Each pseudowire edge device is responsible for maintaining the state of both forward- and reverse-path traffic. Information on the forwarding paths is communicated to the pseudowire edge devices via Forward- or Reverse-Path indication notifications. Table 7-2 discusses the traffic impacts of the received messages.
Table 7-2. Indication Notifications
Impacts ability of the pseudowire edge device to receive traffic over the TDM circuit from the local TDM device.
Note: The pseudowire edge device may be able to detect this directly if the failure occurs in the local port or link.
Impacts the ability of the pseudowire edge device to receive traffic from the remote TDM device
Note: A Forward-Path indication on the PSN does not necessarily imply that the PSN is working improperly, because the defect may be in the remote TDM circuit.
Impacts the ability of the pseudowire edge device to send traffic to the local TDM device.
Impacts the ability of the pseudowire edge device to send traffic to the remote TDM device.
Note: This indication may be indicative of either a PSN fault or a remote TDM fault.
TDMoIP includes its own Operations and Maintenance (OAM) signaling path for reporting of bundle status and performance statistics. OAM signaling provides increased reliability to a protocol stack (TDMoIP pseudowires) that is inherently not reliable. The messages are similar to ICMP messages for the IP network.
Connectivity messages are sent periodically from pseudowire edge to pseudowire edge. A response from the remote pseudowire edge device indicates connectivity. Because forward and receive paths may be different, connectivity messages must be sent in both directions.
Performance messages are sent either periodically or on-demand between pseudowire edge devices. Metrics pertinent to pseudowire performance, such as one-way and round-trip delay, jitter, and packet loss, may be measured.
In addition, standard PSN mechanisms, such as Bidirectional Forwarding Detection (BFD) and MPLS Label Switch Path Ping (LSP-Ping), or other protocol-specific detection mechanisms (L2TP mechanisms described in RFC 3931) may be used over each individual pseudowire, as well as the tunnel itself. These mechanisms may be used continually (proactive notification of defects) or on-demand (reactive notification of diagnostics).
Circuit Emulation Services over Packet-Switched Networks (CESoPSN)
Circuit Emulation Services over Packet-Switched Networks (CESoPSN) is defined in RFC 5086, which was first drafted in January 2004.
Packet structure of CESoPSN is very similar to that of TDMoIP, except for the inclusion of an optional fixed-length RTP header. This packet structure is illustrated in Figure 7-15.
Figure 7-15 CESoPSN Packet Structure
CESoPSN may use an optional RTP header for the transport of timing information. Timing is further discussed later in the chapter in the section, “Timing.” The RTP header includes specific timestamp information that can be retrieved in the following two manners:
- Absolute Mode: In Absolute Mode, the edge pseudowire device recovers the clock information from the incoming TDM circuit. In this mode, the timestamps are closely correlated with sequence numbers.
- Differential Mode: In Differential Mode, the edge pseudowire device has access to a high-quality synchronization source. In this mode, timestamps represent the difference between the synchronization source and the TDM circuit.
CESoPSN Versus TDMoIP
Although both CESoPSN and TDMoIP provide for transport of TDM frames over PSNs using pseudowires, there are numerous differences between the two protocols themselves. These differences include the following:
TDMoIP uses the structure-indication and structure-reassembly mechanisms, whereas CESoPSN uses the structure-locking algorithm. Therefore, CESoPSN transmits consistent, fixed-length packets, whereas TDMoIP has several payload lengths (minimum of 48 bytes) depending on the type of traffic being transmitted.
- This allows for CESoPSN to have a lower packetization delay in instances where the pseudowire is carrying multiple timeslots.
- By the same token, using structure-locking creates inefficiencies when transporting unstructured T1 streams. CESoPSN payload is required to begin at a frame boundary. This means that T1 frames must be padded to create the consistent packet size.
- CESoPSN mandates use of RTP.
By transporting entire frames, CESoPSN simplifies packet loss compensation.
- CESoPSN does not need to look at individual timeslots. Instead, CESoPSN inserts a packet of all 1’s, simulating TDM fault mechanisms.
- TDMoIP must look for structure pointers, jump to the beginning of the next structure, and insert interpolated data.
An ATM pseudowire uses an MPLS network for the transport of ATM cells.
Defined in RFC 4717, ATM pseudowires provide many of the same benefits as TDMoIP and CESoPSN:
- Simplification of network architecture and reduction of number of core networks supported
- Preserving existing legacy services during migration to next-generation IP services
- Using a common PSN to provide both legacy and next-generation services
The generic architecture of an ATM pseudowire service is illustrated in Figure 7-16.
Figure 7-16 ATM Pseudowire Architecture
As with all pseudowire services, the intent of an ATM pseudowire is not to perfectly emulate the traditional service, but instead to provide a transport mechanism for the service. This means there are distinct differences between the traditional ATM service and an ATM pseudowire, namely the following:
- ATM cell ordering is optional.
- ATM QoS model can be emulated, but is application-specific in nature.
- ATM flow control mechanisms are not understood by the MPLS network, and therefore cannot reflect the status of the PSN.
- Control plane support for ATM Switched Virtual Circuits (SVCs), Switched Virtual Paths (SVPs), Soft Permanent Virtual Circuits (SPVCs), and Soft Permanent Virtual Paths (SPVPs) are supported only through vendor-proprietary solutions.
Figure 7-17 illustrates the general encapsulation method for ATM pseudowires.
Figure 7-17 ATM Generic Encapsulation Method
The PSN Transport header is used to transport the encapsulated ATM information across the network. The structure of this header depends on the type of transport protocol being used.
The pseudowire header maps an ATM service to a particular tunnel. If MPLS is being used, for instance, the pseudowire header would be an MPLS label.
The ATM Control Word contains the length of the ATM service payload, sequence number, and other relevant control bits. There are two types of control words that can be used, as follows:
- Generic Control Word: This control word is used for ATM One-to-One cell mode and ATM Adaptation Layer (AAL) 5 Protocol Data Unit (PDU) frame mode.
- Preferred Control Word: This control word is used for ATM N-to-One cell mode and AAL5 Service Data Unit (SDU) frame mode.
Cell Mode Modes
There are two methods for encapsulation of ATM cells: N-to-One mode and One-to-One mode.
N-to-One mode is the only required mode for ATM pseudowires. This encapsulation method maps one or more ATM Virtual Circuit Connections (VCCs) or Virtual Path Connection (VPC) to a single pseudowire. The N-to-One mode allows a service provider to offer an ATM PVC- or SVC-based service across a PSN.
With N-to-One mode, the ATM header is unaltered during this encapsulation, so ATM Virtual Path Identifier (VPI) and Virtual Circuit Identifier (VCI) are present. This information is required to be preserved since concatenation of cells from multiple VCCs may occur.
N-to-One mode has the following limitations:
- Explicit Forward Congestion Indication (EFCI) cannot be translated to a PSN congestion mechanism. Conversely, PSN congestion mechanisms cannot be translated to EFCI.
- Cell header detection/correction that exists in ATM cannot be replicated in the PSN.
- Cell encapsulation only functions for point-to-point MPLS Label Switched Paths (LSPs). Point-to-multipoint and multipoint-to-point are not supported.
One-to-One mode is an optional encapsulation method that maps a single VCC/VPC to a single pseudowire. Because only one VPI/VCI is transported on a pseudowire, the pseudowire context (MPLS Label, for example) is used to derive the corresponding VPI/VCI value. The One-to-One mode also allows a service provider to offer an ATM PVC- or SVC-based service across a PSN.
The same limitations as N-to-One mode apply for One-to-One mode.
AAL5 Frame Encapsulation
There are different optional encapsulation methods that exist specifically for AAL5—one for SDUs and one for PDUs.
AAL5 SDU frame encapsulation is more efficient than using either N-to-One or One-to-One for AAL5. Because the pseudowire edge needs to understand the AAL5 SDU in order to transport it, the device must support segmentation and reassembly.
AAL5 PDU frame encapsulation allows for the entire AAL5 PDU to be encapsulated and transported. Because of this, all necessary ATM parameters are transported as part of the payload. This simplifies the fragmentation operation because all fragments occur at cell boundaries, and the Cyclical Redundancy Check (CRC) from the AAL5 PDU can be used to verify cell integrity.
Figure 7-18 illustrates the four possible locations for defects on the ATM pseudowire service. These four locations are as follows:
- (A): ATM connection from ATM device to pseudowire edge device.
- (B): ATM interface on the pseudowire edge device.
- (C): PSN interface on the pseudowire edge device.
- (D): PSN network.
Figure 7-18 ATM Defect Locations
In all cases, the pseudowire edge device uses standard ATM signaling methods to notify the receiver of cell loss. This information is transported across the PSN to the receiver.
SONET/SDH Circuit Emulation over Packet
To transport SONET/SDH over packet, the Synchronous Payload Envelope (SPE) or virtual tributary (VT) is fragments, prepended with a pseudowire header, and optionally a RTP header. The basic CEP header is illustrated in Figure 7-19.
Figure 7-19 Basic CEP Header
The CEP header supports both a basic mode, which contains the minimum functionality necessary to perform SONET/SDH CEP, and an extended mode, which contains additional capabilities for some optional SONET/SDH fragment formats. These options fall into two categories, as follows:
- Dynamic Bandwidth Allocation (DBA) is an optional mechanism for SPE transmission suppression on a channel-by-channel basis when one of two trigger conditions are met—that the SONET/SDH path or VT is not transmitting valid end-user data or that the circuit has been de-provisioned, or unequipped.
- Service-Specific Payload Formats are special encapsulations that provide different levels of compression depending on the type and amount of user data traffic. The payload compression options are provided for asynchronous T3/E3 Synchronous Transport Signal 1 (STS-1), fractional VC-4, fractional STS-1, and others.
When fragmented, the SONET/SDH fragments must be byte-aligned with the SONET/SDH SPE or VT. That is, the SONET/SDH byte cannot be fragmented, and the first bit in the SONET/SDH must be the most significant bit in the SONET/SDH fragment. In addition, bytes are placed into the fragment in the order in which they are received.
SONET/SDH CEP lies above the physical layer, and assumes that native transport functions, such as physical layer scrambling/unscrambling that SONET/SDH optical interfaces perform as part of their binary coding, occurs as part of the native service. However, CEP does not assume that scrambling has occurred, and fragments are constructed without consideration of this.
Abis/Iub Optimization for GSM Networks
Chapter 2 discusses GSM RAN Abis interface and UMTS RAN Iub interface. GSM RAN Optimization is a method for optimizing and encapsulating structured (NxDS0) TDM signals between the BTS and BSC into IP packets. The optimization is performed by removing nonessential traffic on the GSM Abis interface. Such nonessential traffic includes idle subrates that have a repeating pattern every 20 msec, idle TRAU frames used to keep subrates in-sync for GPRS, and speech TRAU frames with silence used to provide white noise that lets the other party know that the call has not been dropped. In addition, High-Level Data Link Control (HDLC) signaling data flows, which are part of the GSM Radio Link Protocol (RLP), can be optimized by suppressing inter-frame flags.
Optimization is done at the bit level, resulting in no impact to voice quality or data throughput. This bit level optimization makes GSM Optimization radio-vendor independent and radio software version independent. Figure 7-20 illustrates GSM Abis optimization.
Figure 7-20 GSM Abis Optimization