Telecom Convergence (Revised)

Purpose

The purpose of this site is to promote telecom convergence and address issues related to convergence.

What Is Telecom Convergence?

To answer this question the following terms need to be defined:

• Telecommunications (telecom) refers to communications over a distance; communications within a single facility are usually not included under the telecom umbrella;
• Telecom convergence includes transport convergence and the convergence of telecom services;
• Transport convergence means that different types of data are transported through a common network in a similar manner;
• Convergence of telecom services implies that different applications are handled in a similar manner from the point of view of the user, with a common user-to-network interface (UNI) for different applications;
• Data communications (datacom) usually refers to communications between computers or between a computer and a peripheral device; with the advent of convergence, datacom and telecom are no longer distinct.

Benefits of Convergence

The main benefits of telecom convergence are:

• Reduced cost - Many of the costs associated with separate networks, such as the costs of separate equipment and facilities, can be avoided;
• Greater coverage - With a converged network, the full set of services can be provided over the entire network (the union of the network segments) as opposed to full service being limited the intersection of the service-specific networks;
• Easier interoperability - Convergence facilitates interoperability among the networks of different carriers, which further expands the effective coverage area;
• Common user interface - A common user interface and a common access method could be provided for all services.

Transport Convergence

For the most part, voice has been transported via circuit switching and computer data has been transported via packet switching. Data networks, such as the networks that constitute the Internet, have been overlaid on the telephone networks.

With transport convergence, various types of data can be handed in the same way and transported through a common integrated network. One way to achieve transport convergence is to packetize all data as it enters the network and switch the packets using common networks elements. For example, all data could be converted to Asynchronous Transfer Mode (ATM) cells and switched by ATM switches.

Transport convergence is not necessarily visible to the user. User interfaces may be different for two services that employ the same transport mechanism. For example, Frame Relay service may employ ATM transport with frames broken up into ATM cells to be transported through the network. Even though they employ a common transport mechanism, Frame Relay and ATM services would employ different user interfaces.

Convergence of Services

With the convergence of services, different applications would be handled in same way from a customer (user) perspective. For example, an IP router at the customer premise could handle both computer data and voice over IP.

Converged services require broadband access, particularly for video. Ideally, fiber access via fiber-to-the-premise (FTTP) would be provided for all subscribers. However, deploying FTTP involves a very large capital expense, and it will take a long time before FTTP is available everywhere. In the meantime it is feasible to support converged servers using digital subscriber line (DSL) techniques.

Providing voice, video, and data communications over a common network with a common user interface is often referred to as "the triple play." A paper describing how the triple play can be supported over existing telephone lines was recently published in Achieving the Triple Play: Technology and Business Models for Success - Comprehensive Report by the International Engineering Consortium (copyright 2006). Click here to read this paper - http://www.filelodge.com/files/room14/361329/Triple%20Play%20Paper.doc

With converged services, different types of data could be transported differently by the network. For example, time sensitive data, such as voice could be given priority. However, it would still make sense to use a common transport network and common network elements for the most part.

Roots of Convergence

For a long time it has been recognized that integrated communication networks are possible. The Nyquist Sampling Theorem, which was formulated in 1928, provides the theoretical basis for an integrated communication network. Nyquist's theorem states that band-limited signals can be accurately represented by discrete samples as long as the sampling rate is greater than twice the signal bandwidth. If samples are digitized, any practical communication signal can be accurately represented by a finite number of bits per second. In this case, transport refers to functions associated with moving bits through the network.

Sampling at the Nyquist rate (or higher) can result in very high data rates so that in many cases direct transmission of digitized samples is impractical. Shannon's Source Coding Theorem implies that data rates can be reduced below the rates associated with Nyquist sampling and direct digital encoding. This theorem provides the basis for data compression. The Source Coding Theorem and related information theory concepts imply that all types of communication are fundamentally the same and provide the theoretical basis for integrated communication networks. When Shannon's seminal paper was published in 1948, transmission and switching technologies could not support an integrated network. However over the following decades, transmission and switching technologies would improve dramatically so that current technologies are more than adequate to support transport convergence for a wide range of communication applications and services. The industry has been slow to exploit these technological advances and develop truly integrated networks.

Convergence Based on N-ISDN

Narrowband ISDN (N-ISDN) supports the convergence of voice and data communications via transport based on circuit switching and common channel signaling. N-ISDN does not support complete convergence. In particular, it does not support full motion video.

N-ISDN could have been enhanced to support full motion video and other applications. Specifically, the following enhancements would have enabled N-ISDN to be the basis for full convergence:

• Wider bandwidth - More DS0 (64 Kb/s) channels for each subscriber;
• Faster switching - Specifically, dynamically varying the number of DS0 channels assigned to a connection;
• Asymmetric connections - Usually, more DS0 channels assigned in the downstream direction than in the upstream direction.

Enhancing N-ISDN to support full convergence is described in detail in the paper "ISDN and the Internet," which was originally published in Computer Networks - The International Journal of Computer and Telecommunications Networking, Elsevier Science B.V. (copyright 1999). Click on this link to read this paper - http://www.filelodge.com/files/room14/361329/ISDN-Internet%20Paper.doc

Convergence Based On B-ISDN and ATM

Convergence based on enhanced N-ISDN didn't catch on. Instead, convergence based Broadband ISDN (B-ISDN) with Asynchronous Transfer Mode (ATM) as the means for transport was proposed. B-ISDN moved away from the circuit switching/common channel signaling approach of N-ISDN and instead adopted an approach based on small fixed size packets (ATM cells).

The small size of ATM cells (53 bytes) reduces delays and makes ATM better than others approaches for packetized voice. ATM has achieved a certain degree of success in enabling the convergence of voice and data communications. However, the spectacular success of the Internet eclipsed ATM and made convergence based on the Internet Protocol (IP) more attractive.

Convergence Based on IP/MPLS

With IP, routing (forwarding) of packets is based on the destination address in the IP header. Since IP is a connectionless protocol, packets in a data stream can be forwarded independently of each other. This approach is suitable for file transfers, but not very suitable for continuous data streams, such as voice and video data streams.

Label switching (or tag switching) is a technique that can be used to incorporate connection-oriented features into IP networks. With label switching, virtual connections are established through the network and labels (connection identifiers) are assigned. Packets are switched (forwarded) through the network based on the labels rather than on the complete destination address. With Multi-Protocol Label Switching (MPLS), virtual connections called label switched paths (LSPs) are established, and labels are attached to packets to identify particular LSPs. Forwarding of an IP packet is based on the 20 bit label in the MPLS header rather than on the 32 bit IP address (for IP version 4) in the IP header. With MPLS, network capacity can be reserved along the LSP so that the quality of service for a traffic flow can be guaranteed.

MPLS and the development of echo cancellation techniques, enable voice to be transported in IP packets with an acceptable quality of service. With the widespread acceptance of voice over IP, nearly all data will be IP data. Thus, a consensus is forming around convergence based on IP/MPLS. The following sections challenge this consensus.

Convergence Based on Fast Circuit Switching

Although convergence based on IP/MPLS is the approach that is generally favored, it is not the only practical approach for achieving convergence. Convergence can be based on fast circuit switching instead of on packet switching.

A paper describing convergence based on fast circuit switching, which was referred to in the paper as "dynamic channel switching," was originally published in IP Applications and Services 2003: A Comprehensive Report by the International Engineering Consortium (copyright 2002). This paper describes a networking approach that combines the advantages of circuit switching and packet switching in an integrated communications network. With this approach, control signals are separated from the data and transmitted separately in a common control channel. Control and signaling operation can be viewed as a logical extension of Generalized MPLS (GMPLS). A variable number of data channels are dynamically assigned to each connection through the network. Channels are quickly switched in to accommodate the data flow through a connection and switched out when they are no longer needed. With this dynamic channel switching approach, a continuous data stream is not altered as it transverses the network and data channels are efficiently utilized even if the data source is bursty. The proposed approach is compatible with legacy networks and can provide a mechanism for convergence of circuit switching and packet switching. Most of the benefits of the proposed approach can be achieved if only network and server equipment, but not client equipment, is upgraded to support dynamic channel switching. Click here to read this paper - http://www.filelodge.com/files/room14/361329/Convergence%20of%20CS%20&%20PS.doc

Equivalence of Fast Circuit Switching and Connection-Oriented Packet Switching

The previous sections indicate that convergence can be based either on IP/MPLS, which is a form of connection-oriented packet switching, or on fast circuit switching. The question then is: Which is the better approach?

Fast circuit switching and connection-oriented packet switching are two sides of the same coin. Fast circuit switching can closely emulate the characteristics of connection-oriented packet switching, and vice versa. Most communication functions can be efficiently supported by either technique. From a theoretical point of view, network convergence can be based on either fast circuit switching or connection-oriented packet switching. From an implementation perspective, the best approach for convergence involves a tradeoff between efficiency and performance on the one hand and compatibility with legacy networks on the other hand.

Both connection-oriented packet switching and fast circuit switching can be readily implemented in the same network using similar protocols and common equipment. Generalized Multi-Protocol Label Switching (GMPLS) can accommodate connection-oriented packet switching via external labels (MPLS headers attached to packets) and fast circuit switching via implicit labels (time slots containing the data). The key to the proposed converged transport network is a label switch that can handle connection-oriented packet switching and fast circuit switching simultaneously. For the packet switching mode of operation, the routing of data by the label switch is determined on the fly based on information in the packet headers. For the fast circuit switching mode of operation, the switching pattern of the label switch is determined prior to the arrival of the data, based on control information in the GMPLS signaling channel. Otherwise, operation of the label switch is similar for the packet switching and circuit switching modes. To accommodate legacy users and enable interoperability with legacy networks, the proposed label switch could also operate as a conventional router to transport unlabeled packets based on IP addresses.

Using current technologies, network elements that are capable of both the connection-oriented packet switching and fast circuit switching modes of operation can be implemented in a straightforward manner. Initially, transport should be based on IP/MPLS, which is compatible with existing networks. However, fast circuit switching has significant advantages over connection-oriented packet switching. Eventually, fast circuit switching could replace the IP/MPLS approach for most applications. This reduces overhead and enhances security. For a detailed discussion of the equivalence of fast circuit switching and connection-oriented packet switching, click here - http://www.filelodge.com/files/room14/361329/Equivalence%20Paper.doc

Summary

The situation with respect to telecom convergence can be summarized as follow:

• Telecom convergence is not a radical new idea. Its roots go back to the Nyquist Sampling Theorem, which was formulated in 1928;
• Previous attempts at convergence have been unsuccessful, for various reasons;
• We've run out of excuses for not moving forward with telecom convergence;
• Telecom convergence based on IP/MPLS is viable and is the best option for the near term;
• Convergence based on fast circuit switching is also viable and has significant long term advantages compared to the IP/MPLS approach.

Engineering Services

If you need help with issues related to telecom convergence, or if you have trouble downloading a linked document, please e-mail me at kevindemar@verizon.net .