Today's network operators can no longer run their metro boxes using a Sonet/SDH-only mentality. The metro has become a melting pot of services, forcing metro operators and equipment vendors to support T1, Sonet/SDH, ATM, Ethernet, and more.
For the past year, many people have focused on handling these services by performing switching functions at the STS-1 level. However, as more services get packed into metro equipment designs, finer granularity will be demanded at the switching level, pushing designers to move the service provisioning down to the VT1.5 level.
The good news, however, is that adding VT1.5 switching capabilities may not be as hard as it sounds. Through the choice of the right VLSI device, equipment developers can implement VT1.5 switching capabilities through a simple card swap.
In this article, we'll look at how designers can effectively add VT1.5 switching capabilities to a Sonet/SDH box. We'll also explore some of the component- and system-level issues designers will face when implementing VT1.5 switching in a system architecture.
VT1.5's Role
VT1.5 switching is the Sonet/SDH equivalent of T1 cross connection traditionally handled by digital cross-connect systems (DCSs). In a Sonet network, network elements (NEs) are used to transport TDM traffic. These elements multiplex and demultiplex signals into and out of STS-1s or Nx STS-1 payloads, depending on required bandwidth, as well as provide protection switching. However, many NEs do not offer grooming for signals already multiplexed within a STS-1.
The one-to-one relationship between an STS-1 and its payload serves many needs. A typical example of a signal mapped into an STS-1 is a DS3, which may be a clear channel carrying ATM cells. However, if the DS3 is channelized into 28 T1s all must take the same path.
VT1.5 provides the grooming capabilities not available in existing boxes. Twenty-eight T1 signals are mapped into a STS-1 container with individual Sonet pointers and overhead bytes for performance monitoring and alarm reporting. As with DS3 signals, each T1 signal is mapped into a fixed portion of the STS-1 container by means of fixed and variable stuffing. With this mapping, individual T1 signals can be rearranged across a group of STS-1 signals without demapping them to T1s or switching them in a traditional DCS.
VT1.5 switching is also very important when mapping Ethernet traffic over Sonet links. Through VT1.5, metro boxes can use STS-1 or individual containers for mapping T1s that match different Ethernet rates (e.g. seven VT1.5s support 10 Mbps). Fast Ethernet is typically mapped into multiple VT1.5s, not T1s, so a traditional T1 cross-connect cannot be used for such rearrangements.
Figure 1 illustrates two wire centers of a typical network deployed by an ILEC. ADMs are used to collect and distribute synchronous signals, data, and asynchronous signals throughout an access area.
Click here for Figure 1
Figure 1: Diagram highlighting existing (on the left) and emerging (on the right) metro wire center architectures.
The legacy wire center on the left side of Figure 1 contains a DCS, a Class 5 switch, an IP router, and a core Sonet ADM to connect to other wire centers. The DCS grooms T1s, DS3s, and DS0s for the purpose of rearranging mixed traffic (e.g. voice, IP frames, ATM and unequipped T1s) to create homogeneous traffic for efficient transport to the systems providing different services. For example, a STS-1 in an ADM may contain VT1.5s carrying T1s from PBXs, routers, cell sites, and customer premise equipment (CPE). At the wire center these T1s may terminate in a voice switch or be forwarded to a different wire center terminating in ISP equipment, a frame relay switch, or another ADM.
The new wire center on the right-hand side of Figure 1 illustrates features similar to the legacy wire center with the core Sonet ADM replaced or expanded through STS-1 and VT1.5 switching to become a multi-service provisioning platform (MSPP). The MSPP absorbs the functions of the access ADMs plus the DS3 and DS1 cross connection functions of the DACS. This not only reduces the number of transport systems in a wire center, it also simplifies operations complexity for ILECs, speeding service provisioning while lowering capital and operating expenses.
Figure 2 shows a block diagram of a typical core Sonet ADM designed to switch only Nx STS-1 paths. Optical line cards are OC-n with optical to/from electrical conversion, Sonet framing, protection access, STS-1 pointer re-timing, transport overhead termination and STS-1 path overhead monitoring. Separate electrical interfaces for DS3 ports are also provided via DS3 mappers performing path termination of the STS-1s plus DS3 mapping and synchronizing functions.
Click here for Figure 2
Figure 2: Core Sonet ADM with VT1.5 switching.
Interface cards for ATM, packet-over-Sonet (POS), or Ethernet-over-Sonet (EOS) SONET with Nx STS-1 payloads can be provided for the architecture highlighted in Figure 2. All interface cards provide dual STS-1 switch access for redundancy.
Typically switch elements are located on separate cards, which may reside on different shelves, requiring delay compensation and differential driver/receiver pairs with a serial protocol. Many ASSP and ASIC switch interfaces use a basic Sonet/SDH structure as a protocol (A1A2 framing, optional scrambling/de-scrambling for clocking and ones density, H1H2 for payload location and B1 for link error monitoring). Features such as protection switching or communication links may tap unused transport overhead bytes.
Figure 2 also illustrates the easiest-to-upgrade location for establishing VT1.5 switching capability by directly interfacing to the STS-1 switch. A single interface to this switch is provided, placing the VT1.5 switch on the same redundancy plane.
Key VT1.5 Functions
As Figure 2 points out, implementing VT1.5 switching can be done through a simple card insert. What is challenging, however, is choosing the right chips to make VT1.5 switching come to life. Let's take a look at the key function required for VT1.5 switching and look at the key elements that must be delivered at the component level.
To switch multiple VT1.5s, the STS-1 paths that contain them must first be terminated then the VT1.5s from the various STS-1s must be aligned. Unlike DS1 switching which requires framing and slip buffering, VT1.5s locate payloads by pointers; to switch a VT1.5 a new pointer is created (retiming) based on a STS-1 structure common to all VT1.5 to be switched. Once VT1.5s have been retimed, they can be switched in time, usually via buffering in ram, and space. Thus the functions required to perform VT1.5 switching include pointer tracking (to locate the start of the STS-1 payload, possibly supplied by the STS-1 switch), STS-1 path termination, pointer re-timing of the VT1.5 paths, limited VT1.5 performance monitoring and time plus space switching of the VT1.5 containers.
Note that the signals are not demapped and desynchronized to T1s as a DCS may require. Switching native T1s is acceptable for T1s, but will not work for the VT1.5 containers if used to transport Ethernet via virtual concatenation, for example. Once the VT1.5s are switched, they are multiplexed into STS-1s and switched by the STS-1 switch. They can now be sent to a distant ADM or even output through a T1 interface card containing mappers, thus providing direct T1 access.
The transport processor is probably one of the most critical components that a designer needs to consider when building the VT1.5 switch. This device provides the interface to the STS-1 switch, which may involve framing, serdes functions required by the STS-1 switch, as well as the STS-1 path terminating and VT1.5 retiming functions. When choosing a transport processor for a VT1.5 switch card, the design engineer must determine the quantity of VT1.5s that require switching based on access ADMs' VT1.5s and local T1 access ports. For example, if the VT1.5 switch card in Figure 2 provides 772 VT1.5s for T1 ports and STS-1s from the ADMs, 1344 VT1.5 pointer re-timers are needed—one retimer for each direction of transmission. The number of VT1.5s also affects the choice of VT1.5 switch.
When selecting the transport processor (assuming that the STS-1 switch is Sonet/SDH-based), only the basic A1A2 framing, H1H2 pointer, B1 error monitoring and possibly K2 (AIS and RDI) are needed, since the transport processor only interfaces to an STS-1 switch, not an optical line. Access to other Sonet/SDH transport overhead bytes is recommended to easily accommodate features specific to a system architecture (e.g. the J0 byte as a backplane link identification function).
STS-1 path termination features supported in the transport processor should include Sonet overhead bytes C2 (signal label identifying VT mapping), B3 (monitors errors), G1 (for remote defects and error counts) and H4 (locates VT1.5 pointers).
The transport processor physical interface to the STS-1 switch required selection of compatible backplane driver/receiver serdes. For example, a 4 x 622 Mbit/s LVDS interconnect would interface to the STS-1 switch.
An Equally Critical Component
The VT1.5 switch element is an equally critical component. Features of importance are full availability, non-blocking capability, capacity in VT1.5s, ability to perform time and space switching, expandability, and redundancy.
Full availability requires that any inlet VT1.5 can reach any outlet VT1.5. Non-blocking requires that any idle inlet can be connected to a given idle output even if all of the other VT1.5 ports are occupied. Typically, an n x n single stage switch provides this, but going to multiple stages requires expansion of internal paths by up to 2:1.
As with transport processor, the capacity in numbers of VT1.5s is a critical element when choosing a switching element. The capacity of the switch will determine the upper bound for the switch and potentially the entire platform. Since not all the VT1.5s can realistically come and go on single ports, both time and space switching are required. Beginning with the number of VT1.5s, the number of stages and VLSI devices needed to achieve a non-blocking VT1.5 can be determined.
Redundancy and expandability are usually determined by product architecture. Typically the STS-1 switch is implemented with two redundant planes as shown in Figure 2. Here the transport processor and the VT1.5 switch are combined into a single redundant unit that is repeated to match the sts-1 switch redundancy. Expandability can be achieved by populating different quantities of transport processors and VT1.5 switch devices on the switch card or by providing additional transport processors to switch device links to expansion switch cards (not shown).
System Design Considerations
Above, we talked about some of the key component choices designers will face. There are also some system level issues that designers need to keep in mind when building a VT1.5 switch card.
A reduction in capacity to handle duplicate paths is one thing that must be accounted for in the system design. The VT1.5 switch naturally supports testing and diagnostics via loop backs. Virtually concatenated VT1.5s do not require any special treatment by the VT1.5 switch.
On the optical interface cards that connect to the access ADMs (see Figure 1), the STS-1s contain VT1.5s switched by the STS-1 switch to the VT1.5 switch stage for grooming or re-timing. The groomed VT1.5s via the STS-1 switch may be returned to the ADM access interface to continue around the ring, or they may be directed to other interface cards.
In some cases grooming is needed only at the access ADM ring interface cards. Thus VT1.5 functions provided by the switch card in Figure 2 are migrated to the optical interface cards used for access ADM terminations. These new line cards, detailed in Figure 3, have similar optical interfaces but switch interfaces now add STS-1 path termination and VT1.5 pointer re-timers.
Click here for Figure 3
Figure 3: Line cards with added VT1.5 switching support.
In newer line cards, VT1.5 switching is added with the VT1.5s reassembled into new STS-1 paths prior to being sent to the STS-1 switch or reflected back to the optical ports. Alternatively, for 100% VT1.5 switching capacity, the existing STS-1 switch element is replaced by a STS-1 switch with VT1.5 as well as STS-1 capability. Line cards must provide the same functions (less VT1.5 switching) arranged slightly differently (also shown in Figure 3).
Wrap Up
Adding VT1.5 switching to an existing ADM can be done as easily as adding a new VT1.5 switch card. With VT1.5 switching, access ADM and DCS functionality is fully integrated, greatly increasing the life and applications of the ADM.
About the Author
Bill Bartholomay is the chief technology officer of TranSwitch corp. Prior to joining TranSwitch, bill was director of the Shelton Development Center for Ascom Timeplex. Bill received a BSEE from the University of California, Davis campus and an MSEE from the Massachusetts Institute of Technology. He can be reached at billb@txc.com.
