Softswitches Challenge Global Carriers
Posted: 08/2000
Softswitches Challenge Global Carriers
The softswitch, or media gateway controller (MGC), may revolutionize telecommunications services. It will push intelligent function to the edge of the network, enable voice and data convergence, and allow the flexibility of local service creation with the possibility of centralized control.
Through open APIs, it fosters wide-scale innovation of services and can empower customers to define or create their own services.
The modular dismantling of traditional network components and elimination of proprietary software have spawned a new generation of switching hardware contenders, which are not constrained by the proprietary circuit switch mega-manufacturers. They also already address the need to push convergence.
Efforts to integrate voice over DSL (VoDSL) gateways, IADs and multiple access protocols are under way.
However, today’s softswitch has limitations for the international service provider that deploys a truly converged network. The amount of interaction among devices has increased, yet this interaction must be carried over a worldwide network.
Unlike most domestic networks, this comprises expensive and scarce circuits, which at times may be unreliable and possess some inherent latency. This highlights the concern for efficient use and balance between distributed intelligence and centralized control.
New Global Telecom
(www.ngt.com) has requested information with our specific requirements and concerns. While a study is not all-inclusive, we have examined hundreds of pages from a dozen vendors, and we are betting on softswitch technology.
But to provide Tier 1 quality on a global level, development must be made in network architecture, reliability, service control, internetworking and session management.
Architecture
Like many modern telecom carriers, we are replacing our time-division multiplexed (TDM) network with IP- and ATM-based packet technologies, in order to carry voice and data over the same network and because of benefits derived from end-to-end voice packetization.
The next-generation network effectively explodes the switching and processing functions from a central point out to the network’s edge. This stretches the switch matrix and control links across the globe. The switch, which housed all functions in one pre-engineered cabinet, now is de-coupled and distributed.
In next-gen architecture, the tandem switch’s trunk bays are replaced by media gateways at the edge of the network. The tandem switch’s matrix is replaced by the packet-based bearer network, and the tandem switch’s control processors are replaced by the softswitch.
The echo control and digital circuit multiplication equipment (DCME) compression that happened outside the tandem switch is handled in the media gateway.
Replacing the tandem switch’s non-blocking switching matrix with an international data network raises several issues. The softswitch must monitor the state-of-the-bearer network, so it can make intelligent decisions to reroute traffic or apply compression coders/decoders (codecs) when the bearer network is full.
To gain maximum efficiencies in the bearer data network, media gateway and softswitch technology must include methods of dynamically allocating bandwidth by self-learning and monitoring traffic loads. While many domestic carriers have the bandwidth to create an underutilized, non-blocking data network to handle their VoIP traffic, it is not economically feasible internationally. It also defeats the purpose of having a converged, multiservice network.
Reliability
While the TDM switch had a reliability factor (typically 99.999 percent), several elements now must be examined. The reliability of each element is a function of hardware selection and operating system selection. Even if each has a “five 9″reliability, when combined, a reduction of total system reliability is incurred.
Vendors that offer a total solution for next-gen voice networks must establish reliability tests and publish them as a systemwide figure. The reality of a multivendor environment complicates this issue and almost demands that carriers test their service environment.
Service Control
Most international deployments must allow for call and service control functions to be deployed to the network’s edge. The database and service provisioning functions also must have the flexibility for centralized or local control, depending on the application.
The ability for all network elements to appear as one device with one point code–a “virtual tandem”–is desirable. This is important for database administration, service control, call control, telephony signaling and network management.
But how much control and intelligence should be provided to the local softswitch?
A completely centralized approach would require individual messaging for each stage of call setup and teardown. This would create tremendous overhead on an international
network.
The opposite approach dictates softswitches in every region operate solely on its resident database. In this case, the softswitch may not be aware of the state of the intended egress point–possibly several nodes away. In turn, this could limit the intelligence with which it routes, compresses and prioritizes.
The best arrangement involves a compromise where the ingress softswitch has been provided enough information to intelligently package and route the call.
Another debate involves where enhanced service applications reside. While it seems clear that Class 4 services are best served within the softswitch, other services related to Class 5 and enhanced retail features may be served through a separate applications server. This is the same debate that TDM switches have with the transaction capabilities application part/intelligent network (TCAP/IN) model.
The current thought within Internet Engineering Task Forces’
(www.ietf.org) session initiation protocol-telephony service interface
(SIP-TSI) is to house the call control layer in the softswitch and offload the service control layer to an outboard applications server.
Enhanced services most certainly will require the user to remain interactive throughout the call event. This means the applications server must maintain a SIP user-agent to user-agent connection, as well as a media pathway to the media gateway network, to enable tone/voice recognition responses and to play interactive voice response (IVR) sequences.
It also means duplicating some hardware and functions already found in the softswitch and media gateways. A more useful approach for international carriers would be to locate as much service control layer as possible in the softswitch and to use the digital signal processors (DSPs) located in the gateways and access applications servers for instructions on other services as needed.
Architecture scalability concerns go both ways. Based on market demands, one regional location may have dozens of circuits while a main site may have thousands. The ability to support both in a cost-effective manner
is essential.
Internetworking
Standards between the softswitch and the MGC are being developed. H.323 gained initial momentum, but it seems that media gateway control (MEGACO) and SIP now are in the spotlight. Protocols also are required among the softswitch, applications server and other softswitches.
Any international implementation needs to address the existing capabilities in many SS7-oriented standards related to routing, e.g., E.170, E.351 E.411 and E.412.
These deal with operational issues such as real-time surveillance, transmission failures, circuit types, signaling conditions, congestion and blocking. Commercial issues include call and network control mechanisms and routing schemes.
When considering a multisoftswitch network or internetworking between carriers, it is imperative to interoperate on the bearer control layer and the call control layer. Otherwise it hinders the quality of international voice due to issues such as multihop voice/fax compression, end-to-end compression algorithms and QoS adherence. This could be accomplished by the global adoption of an SS7 variant such as Bearer Independent Call Control (BICC) (ISDN user part, ISUP+), H.323, or other packet-based protocols.
Once all these standards are in place, we need the capibility to modify them. Control through using Layer 3 message transfer part (MTP) and Layer 4 ISUP screening and field manipulation allows carriers to compensate for nonstandard connectivity.
Routing Capabilities
Most current routing schemes do not support the complex nature of the international internetworking. Carriers will continue to connect with other foreign and domestic carriers to achieve cost-effective global coverage.
But they define their network services and costs differently. What is a cellular call to one carrier may not be to another. And cost may differ by a factor of three.
Session Management
Finally, we must have session management reporting. Individual elements need to report call events to be correlated by network and service management systems.
The International Softswitch Consortium
(www.softswitch.org) session management working group recommends interoperability of upstream network management and mediation systems within networks containing independently developed switching, gateway and border element devices through standard format.
Numerous efforts on accelerated tracks are being made to address these issues, but many developments diverge. Also, an effort has begun to adapt this technology to the markets with mature infrastructure and rapid growth in available bandwidth.
The international provider will require some special development to reap these technology benefits. The effects on their economies should be dramatic.
It simply is not enough to design a network that works. You must build one that works with other diverse networks and keeps working through unexpected conditions and global changes.
Lee Story is senior vice president of operations
and technology (lstory@ngtele.com), Mark Bolton
is director of research and development (mark@ngtele.com) and Jon Hickam is a network architect
(hickam@ngtele.com). All three are designing and implementing New Global Telecom’s
(www.ngt.com) next-generation network.
By Lee Story, Mark Bolton and Jon HickamPHONE+ invites you
to air your views.
Call us at +1 480 990 1101
or e-mail khenderson@vpico.com.
ROUNDTABLE
On who’s slamming whom …”AT&T has a history of slamming violations in long distance. Now they have transferred their illegal slamming activities into the local toll market in California.”
–Sandy
McGreevy, director of consumer protection, Pacific Bell (www.pacbell.com)
“SBC/Pacific Bell’s [slamming] actions demonstrate that it is not willing to compete on a level playing field.”
–Lois
Hedgpeth, president, AT&T Pacific Region (www.att.com)
On June 6, WorldCom Inc. (www.wcom.com) “agreed to pay the largest slamming payment in the history of the commission and to substantially change their behavior toward their customers. This action puts the entire industry on notice that consumers should not be hoodwinked over the phone. We have zero tolerance for slammers.”
—William E. Kennard, chairman, FCC
(www.fcc.gov)
“The FCC and the states will be stepping up their information sharing of various specific company slamming activities. As a result, repeat offenders will be more likely to suffer additional penalties, including the potential loss of an operating certificate in a particular state or an additional fine for multiple offenses imposed by the FCC.”
—Bob Rowe, president, National Association of Regulatory Utility Commissioners
(www.naruc.org)
