×

SUBSCRIBE TO TMCnet
TMCnet - World's Largest Communications and Technology Community

CHANNEL BY TOPICS


QUICK LINKS




 

FeatureArticle.gif (4903 bytes)
October 1999


Balancing Voice And Quality On The LAN

BY KEN DUMONT

Today many vendors, including well-established vendors and startups, are introducing LAN-PBX solutions. These products do not measure up to the voice quality and reliability standards offered by current PBX-based telephone systems. Customers need to understand the limitations when implementing a telephone system based on current LAN-PBX technology.

The bar for acceptable voice system performance has been established, and enterprises will not expect anything less from a LAN-PBX solution. Moving to a LAN-PBX architecture should not force customers to compromise on what they have today. As a result, there are several key issues that need to be addressed before solutions can meet and exceed the expectations set by existing products, including reliability, voice quality, interoperability, usability, and management.

HOW CAN A LAN-PBX BE MADE RELIABLE?
Businesses understand that a LAN-based telephone solution needs to be as reliable as their current installed PBX solution. For most enterprises, telephony represents the original “mission critical” application, implemented long before other data applications became critical.

One way to achieve high reliability is to install redundant and fault-tolerant system components. Redundancy provides fault tolerance because if a component fails, the hot spare will take over without any loss of service. Further, fault tolerance can be achieved through the use of hot-swappable components. Even if a component fails, a truly reliable product will allow the technician to replace the faulty unit without powering down the entire system. These types of fault tolerant and redundant equipment techniques are expected in today’s PBXs, but are not necessarily included in PC servers, including any PC-based server that runs call server software in a LAN-PBX architecture. While many of the server’s components can be made redundant, this reliability comes at a cost that many small and medium-sized businesses may not be able to afford.

Also, PC servers that run multiple applications will be exposed to more failures due to the “misbehavior” of other applications. It is typically recommended that the organization run the call server software on a dedicated hardware server, at least for the near term.

In addition to hardware failure, server downtime can also be attributed to power loss. There are several Uninterruptible Power Supply (UPS) products on the market that will keep PC servers running for a limited amount of time. Unfortunately, most of these products are not designed for high server availability in the event of a significant power failure, but are only meant to allow users to save their current data and allow the administrator to gracefully shut down the server. This is not acceptable for a telephone system, so administrators will have to implement a battery backup system for all LAN PBX components — a system that can maintain power to the call server, network connectivity devices (such as Ethernet switches and routers), and Ethernet phones in the event of a power outage. All of this adds to the cost and complexity of the voice and data network.

CAN TOLL QUALITY VOICE MEASURE UP?
A common belief is that most LANs have enough excess capacity to handle the relatively small amount of traffic generated by voice applications. However, traffic loads are dynamic on a LAN, and only the use of specialized and sophisticated bandwidth allocation techniques can guarantee available bandwidth at all times of operation. These factors, coupled with the lower bandwidth of WAN connections, mean that LAN-PBX systems must be prepared to deal with varying bandwidths without sacrificing quality.

The bane of any packetized voice stream is delay. Once a packet takes longer than 100–200 ms to travel to its location, voice quality starts to become unacceptable. Human factors’ research has shown that most people greatly dislike long transmission delays in telephone calls. Subjectively, too much latency makes it difficult to carry on an interactive dialog, which is typically the case in voice communications. This is particularly apparent when people try to interject comments or questions while someone else is talking. One party hears that the other has stopped speaking and begins to talk, only to find that the second party merely paused before continuing to say more. As latency increases, people tend to “step all over each other” in the conversation, not unlike collisions on an Ethernet LAN.

Other factors that affect voice quality in a LAN-PBX system are echo and packet loss. Echo becomes a problem in most voice-over-packet networks be-cause the round-trip delay through the network is almost always greater than 50 ms — the threshold of human perception. These systems must address the need for echo control, and may need to implement some means of echo cancellation.

INVESTMENT PROTECTION AND LEGACY MIGRATION
To protect your current investment in legacy voice systems, a LAN-PBX needs to be compatible with the existing infrastructure. To date, vendors are not selling their products to companies with existing PBXs, but to smaller companies that have no PBX and have outgrown their key systems or Centrex service.

First, if the PBX has adequate capacity, and users and administrators know how to program it, why settle for a voice and data solution that has fewer features, and different implementations of these features? Second, switching to a LAN-PBX system may represent a “fork-lift” upgrade that requires all new phones and some re-wiring. Third, the average life span of a telephone system is shrinking, but still typically five to 10 years (compared to 18 months to three years for a PC). Finally, companies have a significant “soft investment” in training users and administrators on how to utilize and administer it. The thought of training users on a new system with new desktop interfaces, new ways of implementing features, or even doing without the key functionality to which they’ve become accustomed, is not particularly appealing.

Thus, any organization considering an implementation of LAN-PBX systems will need to consider asset protection for its pre-existing systems (and people). Ideally, a LAN-PBX system should work in a “hybrid” implementation along with existing legacy proprietary systems to protect both hard and soft voice investments. Examples are branch offices that are networked to the head office, or departmental adjuncts within larger organizations. Hybrid implementations need to deliver seamless integration, feature and functionality transparency between the IP and traditional systems, common telephone set interfaces, etc. Also, vendors need to demonstrate a clear migration path between legacy voice equipment and next-generation platforms. This will define the product roadmap, which products will migrate forward (and which will not), and how legacy systems will interoperate with new platforms.

MANAGING A LAN-PBX
PBX telephone systems are typically managed by a physical port — each port can be assigned a phone number, name, location, and calling access and restrictions. This management and configuration is performed using a graphical interface that communicates directly to the telephone system hardware.

With a LAN-PBX system, there is no concept of a PBX port. Instead, administration will need to be associated with a particular user name or some other differentiation, such as the MAC or IP address of an Ethernet phone.

Ideally, a LAN-PBX system would be managed based on uniquely defined user names. With such a system, a user would have the same telephone number and access rights even if he/she was using a different Ethernet phone. By tying this information into a directory service — potentially based on Directory Enabled Networks (DEN) or Lightweight Directory Access Protocol (LDAP) — the administrator would be able to globally manage every Ethernet phone in the entire enterprise from a central location.

Another issue that is unique to a LAN-PBX system is policy-based management — maintenance of access control lists and quality-of-service settings per user or Ethernet phone. As phones move onto the data network, the same management techniques that are used for controlling PCs and users will need to be implemented for LAN-PBX components and users.

CHALLENGES
The network and telecommunications equipment industry faces significant challenges in driving widespread adoption of LAN-PBX solutions, but the momentum towards IP-based network convergence is inevitable. Proof of the value of voice/data convergence in the wide area will eventually cause it to happen in the enterprise. It’s a matter of when, not “if” LAN-PBXs will become popular.

In order for LAN-PBX solutions to compete on an equal footing with traditional PBXs, and be accepted by the mainstream market, they must address the practical challenges, including reliability, voice quality, interoperability, and management. c

Ken Dumont is vice president of marketing for Mitel Corporation. Mitel’s product portfolio has evolved from world leading business communications systems and desktop peripherals to include networked voice and data systems, a portfolio of call center solutions, messaging products, computer telephony (CT) applications and enabling platforms, network access products, remote access products, and public switching products. For additional information, visit Mitel’s Web site at www.mitel.com.


The Intelligent Network For VoIP

BY KRISHNA MURTI

For many individuals, the Intelligent Network (IN) is synonymous with the SS7 signaling network and the robust databases that enable advanced telephony services. Over the past 25 years, the IN has evolved to become an almost ubiquitous network that is a storehouse for subscriber data, routing tables, service logic, billing data, and other information that enables service providers to deliver advanced services. In fact, even today’s basic telephony services rely extensively on the IN for call completion. Although the IN is a separate, parallel network, it has become an integral part of the current circuit-switched network. It plays a fundamental role in service delivery, and provides much of the intelligence in the Public Switched Telephone Network (PSTN). How then, can we leverage this capability to support service delivery in the emerging IP telephony market?

Before exploring IN for Voice over IP (VoIP), it’s useful to first review the network architecture of the current PSTN. Within the central office, a Class 5 switch provides subscriber line access, circuit switching, basic call control functions, and interoffice trunking of voice traffic. Signaling among central offices is usually transported through the highly reliable SS7-based packet network that is part of the IN infrastructure. The packet switches used for transport of signaling are called signaling transfer points or STPs. Other IN elements, specifically, the service control point (SCP), service management system (SMS), and service control environment (SCE) provide much of the “intelligence” in the IN. It is these latter elements, in conjunction with the Class 5 switch, that play a part in delivering advanced telephony features such as number translation (800/888 toll-free, local-number portability), calling card services (prepaid as well as credit), caller ID, call waiting, call screening, and a variety of other services. The above architecture, or a slight variant, is used by many carriers to deliver seamless, reliable telephony services for both wireline and wireless networks.

As the communications industry gears up to support Internet telephony, the traditional central office is undergoing dramatic changes. Many functions that used to reside in a single monolithic switch are being distributed in a variety of gateways and servers, and a new element, often called a “softswitch,” is emerging. Depending on the specific signaling protocols being supported, these softswitches are sometimes referred to as gatekeepers or call agents. Just as the Class 5 switch handled signaling and call control in the circuit network, softswitches provide these functions in packet networks.

As service providers deploy Internet telephony, many would like to continue to leverage the IN. This is particularly true for carriers that already have significant IN investment. Often, these carriers have spent many years and tens or even hundreds of millions of dollars in building their IN infrastructure. A service provider’s IN typically houses valuable subscriber data and complex service logic that would be difficult or at least time-consuming to replicate. Rather than throw away this asset and start anew, traditional service providers are looking for ways to reuse their existing IN to support telephony services in a packet environment.

An alternative scenario, which appears particularly attractive to next-gen telcos or service providers without an IN infrastructure, is also feasible. Instead of relying on IN-SCPs, these service providers plan to deploy distributed servers and directories that contain the required databases and service logic to support advanced services. This scenario appears to overcome some of the objections raised by opponents of the traditional IN approach — high cost of infrastructure, and a relatively complex and tedious process for developing and deploying new services. Other proponents of a non-traditional IN model envisage a highly decentralized network with intelligence broadly distributed in servers and in end points. It is likely that a variety of alternative approaches will emerge, some of which may be viable.

Organizations such as the International Telecommunications Union (ITU), the International Engineering Task Force (IETF) and ETSI TIPHON, as well as various ad hoc industry groups, are working hard to standardize protocols, interfaces, and network architectures for the emerging VoIP market. Significant effort is being expended to support IN interworking in these environments, specifically to enable the transport of SS7 protocols such as ISUP and TCAP over IP. There is also active work on directory-enabled networks, new call processing languages, service creation environments such as JAIN, PARLAY, and OSPA, and a variety of technologies (e.g., VXML) that will enable new services.

The industry is undergoing radical change and many new ideas will emerge. In the short term, the existing IN is a valuable asset. Service providers with the resources to leverage this infrastructure will be able to rapidly deliver many of the existing voice services. Those who opt to ignore the IN will need to forgo existing services or will spend a significant amount of time recreating those services using a different architecture. A likely scenario is that many service providers will leverage the IN to support existing services, and over time, will rely on alternative approaches for new services.

Krishna Murti is CTO of elemedia. elemedia, a wholly-owned venture of Lucent Technologies, is a leading supplier of a complete suite of standards-based, carrier-grade software platforms for OEMs and service providers targeting multimedia communications over IP networks. For more information on elemedia, visit their Web site at www.elemedia.com.







Technology Marketing Corporation

2 Trap Falls Road Suite 106, Shelton, CT 06484 USA
Ph: +1-203-852-6800, 800-243-6002

General comments: [email protected].
Comments about this site: [email protected].

STAY CURRENT YOUR WAY

© 2023 Technology Marketing Corporation. All rights reserved | Privacy Policy