A network—if built—would be able to carry all types of services and protocols.
A group of Russian telecommunications scientists has developed a new technology that can serve as a backbone for today’s multiple communications protocols or as a stand-alone network. It can be scaled from a local area network up to a global telecommunications system capable of carrying voice, data and video simultaneously.
The St. Petersburg, Russia-based researchers developed the technology to serve all types of telecommunications traffic without many of the problems that limit current systems attempting to carry various forms of voice, data and video communications. Instead of finding ways to piece together different modes and protocols, the scientists came up with a system that can serve as a foundation for diverse systems or carry all kinds of traffic on its own.
The single biggest obstacle to building a vast network based on this innovation may not be technical but financial. Cash-strapped Russian telecommunications vendors cannot yet afford the large expenditures necessary to proceed with the next step in the technology’s development, let alone build a new nationwide network across the breadth of the world’s largest country.
Known as synchronous asynchronous transfer mode, or SATM, the new technology was developed by researchers at the State University of Telecommunications in St. Petersburg. Dr. Vladimir K. Kharitonov, head of the SATM development team, states that SATM can carry multifaceted telecommunications services currently in demand while eliminating many of the drawbacks that plague other technologies and protocols.
SATM combines simplicity and power, quality-of-service guarantees and perfect bandwidth utilization, scalability and easy integration with existing network technologies, he states. “SATM technology can be considered as a simple and natural basement for creating next-generation multiservice networks.”
Kharitonov’s research that led to SATM focused on uniting the key benefits of traditional telephone networks—circuit switching and its guaranteed native quality-of-service support—with the benefits of data networks—good bandwidth utilization, low overhead and asynchronous multiplexing. This effort was driven by the reality that current telecommunications modes did not perform well when they tried to cross disciplines.
“Telephone networks were developed for voice transmission, and data networks were developed for data transfer, but not for multimedia traffic,” Kharitonov continues. “That is why so many problems arise when, for example, one tries to adapt IP [Internet protocol] for transferring voice or video.
“None of the existing technologies can serve as the basement of a multiservice network,” he declares.
Kharitonov views several criteria as necessary for next-generation telecommunications networks. These include a simple structure; minimum signaling information at the time a connection is completed; native support of different classes of service, combining strict and soft resource reservation, synchronous and asynchronous multiplexing; transparent transmission of existing network traffic; and support for new services. He emphasizes that, while SATM was developed to serve as a global end-to-end network, it also can serve as a backbone to transfer other networks’ traffic, even as a local area network.
Kharitonov explains that SATM’s switching technology includes features of both circuit and packet switching, which allows all types of traffic to be transmitted in their native formats. It uses circuit switching for real-time traffic with constant speed, and it uses statistical multiplexing for variable-speed real-time traffic. Asynchronous features are employed for data traffic. Unlike transfer control protocol/Internet protocol (TCP/IP) or asynchronous transfer mode (ATM), SATM requires no emulation.
He notes that ATM’s cell switching technology was developed to eliminate the shortcomings of packet switching and to transmit different kinds of traffic. ATM’s main feature is its short, fixed-length packets and high transmission speed. However, these packets require headers, and their overhead takes up about 10 percent of each packet’s 53-byte capacity. In addition, ATM hardware is complex and expensive, and it requires high-speed channels.
In contrast, SATM does not require headers for its highest-capacity traffic. For some data traffic, it may require 1 control bit per data block. With its varied block sizes giving the technology additional flexibility, the overhead for a 16-byte data block would be less than 1 percent. SATM also uses simple and native quality-of-service mechanisms that perform as in a telephone network. And, SATM’s synchronous-asynchronous multiplexing allows it to work even on 64-kilobit-per-second access network channels. Data blocks are multiplexed within a limited time interval.
Switching delays would be significantly reduced, as SATM introduces statistical multiplexing within the limited time interval. Switching delay is determined by frame duration, which in a backbone SATM network is 125 microseconds. Kharitonov offers that SATM switching operations are comparable in simplicity to those of a standard telephone network.
Another key advantage is that SATM does not use a complex adaptation layer. This makes it transparent for existing network technologies such as TCP/IP, whereas ATM with its complex adaptation layer is more difficult to integrate with TCP/IP.
A user would have little trouble sending IP over SATM. Kharitonov explains that IP could be used as a signal and protocol to establish a connection over SATM. For quality of service, SATM would handle the actual traffic.
Because SATM supports transmission speeds ranging from kilobits per second up to gigabits per second, its developers have designed six different classes of service geared toward various types of traffic. Four of these classes are similar to ATM classes, to which two other new types are added.
Class 1 transmits traffic that requires constant speed. This might include video or 64-bit-per-second voice. Kharitonov notes that this class of service is based on synchronous circuit switching, which involves strict resource allocation. In contrast, ATM Class A is based on asynchronous transfer, which can feature jitter and information loss.
Class 2 supports real-time traffic with variable transmission speeds, such as for video and sound streams. It provides this transmission without loss or jitter. As with Class 1, it uses strict resource reservation to guarantee transmission quality.
Both Class 3 and Class 4 support transmission of variable speed traffic. Unlike classes 1 and 2, these two classes allow jitter. Class 3, which is similar to ATM Class B, allows cell losses during network overloads, so it is better suited to low-demand real-time traffic. Class 4, which features constant-size data blocks, does not allow losses. However, it is buffered, so it allows more jitter than Class 3.
Class 5 permits variable-size data blocks, as it does not have any strict requirements for size or delay variation. Its packets are marked with labels that identify the type of Class 5 connection.
Class 6 is designed for connectionless data transmission in the same manner as TCP/IP traffic. Each packet carries information about source and destination node addresses.
Customers would select the appropriate class for their transmissions. Kharitonov explains that a user sending video, such as in a videoconference, might opt for Class 1 or Class 2. Conversely, a simple data transmission such as an e-mail message might be sent Class 6. Customers would pay different rates for each class, with Class 1 being the most expensive.
Different classes of traffic can be mixed in transmissions. For example, an SATM Class 1-4 switch performs switching for that class of traffic while it demultiplexes Class 6 traffic from frames and sends it to an IP router, which performs the Class 6 routing. The IP router forwards packets to another port of an SATM switch, which handles the packets received from the IP router as Class 6 traffic. It then forwards them to an appropriate SATM port, where they are multiplexed back with Class 1 and Class 4 traffic. The IP router itself could be integrated into an SATM switch.
A country employing any telecommunications infrastructure, either older or state-of-the-art, can easily incorporate SATM technology, Kharitonov states. SATM can be used as a backbone to transfer other networks’ traffic, or it can be an end-to-end network. It offers transparent integration with both telephone and data networks, so an existing telecommunications infrastructure need not be replaced concurrent with the introduction of SATM. This would permit some network customers to use old-style network services, while other customers simultaneously could use SATM services.
Several technological challenges arose during the first year of SATM development, Kharitonov allows, but these largely were eliminated after several months. For example, the first version employed a fixed unit size, but this proved to be a limitation because it did not permit SATM to support slow speeds such as 8 kilobits per second. Also, creating a variable-size control field required synchronizing different equipment when a connection was established, so the current version of SATM uses a fixed-size control field in the beginning of each SATM frame. Accordingly, SATM switches can be synchronized easily at the time a connection is established. The first SATM version also lacked different classes of service.
Kharitonov does not hesitate to say that financial constraints are the biggest challenge facing deployment of an SATM network. None of Russia’s nascent telecommunications vendors has the resources to support SATM development.
While the first stage of SATM development is completed, the second stage involves development of SATM standard low-level specifications. Developing these specifications will require considerable investment from a large telecommunications company or a consortium of experienced firms, Kharitonov warrants. Accordingly, his team is actively seeking commercial partners in this vital next step.
Additional information on synchronous asynchronous transfer mode is available on the World Wide Web at www.satm-tech.com.