The explosive growth of the Internet has dramatically increased the demand for a modem that exceeds V.34 and V.34+ speeds. With the new transmission schemes, modems deliver line rates of up to 56 kilobits per second (kb/s), nearly double the line rate currently supported.
Today, 56k modem products are available from several manufacturers, with many more to follow in the coming months. A test procedure to evaluate the performance of these new products in the presence of current telephone network conditions must be used to accurately characterize 56k modem performance.
The Basics of 56k Modems
Today’s modems provide line rates above 30,000 b/s and, through the use of data compression techniques, data transmission rates in excess of 100,000 b/s. Until now, industry experts felt that the current V.34 and V.34+ modem designs, which supported data rates up to 33,600 b/s, had stretched the telephone network to its theoretical limit. However, the modem industry has seemingly broken through that barrier with the announcement of the 56k modem.
The development and use of the 56k modem have become feasible primarily because digital transmission facilities have been rapidly replacing analog transmission facilities on the Public Switched Telephone Network (PSTN). Figure 1 illustrates the typical network configuration for a connection between two 56k modems.
The central site modem transmits 8-bit data words over digital transmission facilities to a pulse code modulation (PCM) codec (coder/decoder) located at the line card. The codec converts the digital data word, representing one of 255 nonuniformly spaced points, to its corresponding analog voltage level using either a mu-law (United States and Asia) or A-law (Europe) coding technique.
The codec processes the 8-bit samples at a rate of 8,000 times per second, producing a theoretical data rate of 64,000 b/s. Network signaling techniques, such as robbed-bit signaling and the inherent characteristics of the codec conversion, limit the maximum data rate to approximately 56 kb/s. Robbed-bit signaling is a technique used on digital transmission facilities on the telephone network to indicate call status, such as on/off hook or ringing. The least significant bit of a digital data word (8 bits) is “robbed” and replaced by a signaling bit.
The subscriber modem transmits and receives data in an analog format via the local subscriber loop. The subscriber modem uses clock recovery techniques to synchronize its sampling clock to the codec clock. This allows the subscriber modem to interpret the appropriate voltage level from the 8-bit sample.
Data must remain in a digital format between the central site modem and the codec for two modems to connect using the 56k technology. Any digital/analog conversion of the signal before it reaches the codec limits the speed of the connection. Digital coding schemes, such as adaptive differential pulse code modulation (ADPCM) and line shaping techniques used on the telephone network, also prevent connections at the higher line rates. ADPCM is a technique that converts an analog signal to digital by taking 8-bit samples at 8,000 Hz. Then ADPCM further compresses the information on the 8 bits to even fewer bits, such as 4 or 5 bits, to increase bandwidth on the telephone network.
Line shaping techniques such as AT&T True Voice attempt to improve voice quality by boosting certain components of a signal and attenuating other components.
Figure 2 illustrates the typical test bench for V.34 and V.34+ subscriber modems. The telephone network emulator generates PSTN transmission impairments such as noise, intermodulation distortion, and echo. The loop emulator recreates the characteristics of the local subscriber loop that interfaces each modem to the telephone network. The dual terminal emulator configures the modems under test, performs call setup, and executes modem performance tests.
There are several key differences in the 56k modem test bench (Figure 3). Since the central site modem connects to the PSTN via a digital interface, the telephone network emulator must provide digital access for the central site modem.
Loop emulation only is required on the subscriber modem side of the test bench. Since 56k modems function as V.34+ modems in a subscriber-to-subscriber configuration, the 56k modem test bench also must provide the functionality of the V.34+ modem test bench shown in Figure 2.
The telephone network emulator in the 56k modem test bench provides characteristics associated with digital transmission facilities. Digital transmission facility elements, such as digital loss pads and robbed-bit signaling, always have been present on the telephone network but had less of an effect on modem performance. These network elements, however, now are of greater interest because they directly affect the digital data words transmitted by the 56k modem.
A digital loss pad is a network element that attenuates a digital signal by mapping digital input words (8 bits) to digital output words (8 bits). Digital loss pads introduce distortion in addition to attenuation and can adversely affect 56k modem performance.
Accurate, controllable emulation of the codec characteristics also must be provided because they can greatly impact 56k modem performance. Codec characteristics include the transmit-and-receive filter frequency responses, the type of codec conversion technique, and quantization distortion. Programmable control of the codec characteristics is vital because the characteristics vary widely depending on the manufacturer of the codec.
The telephone network emulator also includes analog transmission facility impairments such as trans-hybrid loss, attenuation, nonlinear distortion, and noise. Trans-hybrid loss is characteristic of the 4-wire-to-2-wire conversion that occurs on the telephone network at the line card.
The loop emulator provides the conditions found on the local subscriber loop, including loop length and frequency-response characteristics. The effects of local loop conditions such as loop noise and low-frequency distortion on 56k modem performance must be characterized to be added to the test bench.
Analyzing Modem Performance
File transfer throughput and call connect reliability tests best measure real-world performance of modems. Modem throughput should be measured using real-world data file types, including graphics, text, or spreadsheet files.
File transfer tests should be performed over a variety of impairment conditions to introduce transmission errors. These tests examine the robustness of the modem design, the data compression algorithm, and the error-correction algorithm, all factors which directly impact throughput.
Call connect reliability tests analyze a modem’s capability to negotiate and establish connections in the presence of typical impairment conditions. The call connect test determines the speed and reliability a 56k modem provides by measuring the connected line rate and the connection success percentage.
In addition, the call connect reliability test examines interoperability between the products of various manufacturers. Interoperability is a critical issue for 56k modems because multiple vendors supply central site and subscriber modems.
56k Modem Test Systems
The Telephone Network Emulator provides accurate emulation of telephone network characteristics, which is the most critical component of a 56k modem test system. The telephone network emulator contains a digital interface for the central site modem and an analog interface for the subscriber modem. Calls originating from either station must be handled by the network emulator.
Digital facility emulation (DFE) provides programmable control of the loss and distortion associated with digital pads. DFE also provides the flexibility to model multiple PCM codec implementations to measure 56k modem performance over a range of typical network conditions. By programming the coding scheme and the transmit-and-receive filter characteristics used by the codec, the DFE can be customized to emulate virtually any type of codec.
A complete, automatic modem test system combines the telephone network emulator with other test instruments and software to provide automated performance testing of 56k modems. These systems use data-analysis software to perform call setup, throughput, and call connect reliability tests on all types of 56k modems. Tests have shown that 56k modems connect at rates below the maximum 56,000 b/s with typical connect rates between 30,000 and 48,000 b/s.
Automatic test software allows tests to be performed over different impairment scenarios in an automated fashion. The system supports all of the required data interfaces and network interfaces to completely test both central site and subscriber modems.
Testing Standards
The TIA TR30.1 Subcommittee is working toward a standard that defines a single 56k modem modulation scheme. Current 56k modem products are based on one of two competitive designs. These modulation techniques currently do not interoperate, which means that central site modems and subscriber modems must use the same technology to achieve a 56k connection.
The TIA TR30.3 Subcommittee is developing new testing specifications that define a network model and standard test procedures for 56k modems. Since 56k modems operate over a different network configuration than V.34 modems and new impairments have been identified, the TR30.3 Subcommittee must create a network model that accurately reflects these conditions. A companion specification that defines the types of tests to perform on 56k modems also is in development.
Conclusions
As the demand for faster data communications continues to grow, the need will rapidly increase for a modem that provides reliable data transmission at higher line rates than V.34+. Modem developers and manufacturers are producing 56k modems to satisfy high-speed data-communications requirements. A complete and accurate test plan to precisely evaluate 56k modems is a critical factor in comparing the performance and capabilities of various products.
About the Author
Michael Pellegrini is the product manager for modem test systems at Telecom Analysis Systems. He joined the company in 1993 and, prior to his promotion in 1994, was the western region sales engineer. Mr. Pellegrini graduated from Drexel University with a bachelor’s degree in electrical engineering. Telecom Analysis Systems, 34 Industrial Way East, Eatontown, NJ 07724, (732) 544-8700.
Copyright 1998 Nelson Publishing Inc.
January 1998
|
|