CMTS Testing Keeps the Data Flowing

In addition to transferring data, a CMTS also manages the cable modems it serves, providing them with timing information and configuration setups and scheduling upstream data transmissions. Figure 1 shows a typical CMTS installation in an MSO’s cable plant. The CMTS, through optical transmitters and receivers, sends data over dozens of kilometers of fiber-optic cable originating at the MSO’s facility, ending with 1 km to 2 km of coaxial cable in a neighborhood.
   

A CMTS must be tested against industry specifications and for how well it works with any manufacturer’s cable modem. CMTS manufacturers typically have in-house test labs with areas dedicated to each type of testing; a manufacturer will run dozens of tests that require hundreds of measurements. These tests require a significant investment in test equipment and materials, with costs running up to several hundred thousand dollars.

In the test lab, engineers perform RF measurements on the communications physical layer and data measurements at the MAC (media access control) layer. These measurements that test equipment perform in a lab help engineers decide if a CMTS is ready for qualification tests at CableLabs (Louisville, CO), an industry nonprofit test lab. CableLabs performs tests based on the DOCSIS set of specifications.¹ CableLabs will certify that a cable modem complies with DOCSIS. In contrast, CableLabs will qualify a CMTS. Qualification tests follow a looser, somewhat subjective, set of guidelines, but they still ensure that a CMTS operates within reasonable parameters.

In addition to performing prequalification tests, CMTS manufacturers run performance tests. Performance tests consist of two main parts: interoperability tests and stress tests. These tests prove that a CMTS will work with hundreds of cable modems made by numerous manufacturers while passing its full-rated load of data traffic. A CMTS must also prove its worthiness under simulated real-world conditions, which include long cable runs with wideband noise and other impairments that can interfere with data reception in either the upstream or downstream direction.

What Are the Tests?
The majority of the DOCSIS measurements occur at the physical layer. Here, engineers measure the RF characteristics of the downstream signal from the CMTS, and they measure how well the CMTS can receive upstream signals from cable modems. At the MAC layer, engineers measure how well a CMTS can identify and configure a cable modem. (See “Range and Register,”, for a description of the process that takes place when a subscriber connects a cable modem to an MSO’s network.)

CMTS downstream RF tests include tests for

• EVM (error vector magnitude). EVM gives engineers an overall figure for noise and distortion. It is the rms difference (expressed in percent) between the I/Q constellation of a received signal and that of a perfect signal. Engineers perform EVM measurements on a demodulated downstream signal.

• Output power level. At the CMTS, this level should range from +50 dBmV to +61 dBmV into a 75-V load.

• Phase noise. The phase noise should measure less than –33 dBc from 1 kHz to 120 kHz from a downstream channel’s center frequency and should measure less than –51 dBc from 10 kHz to 3 MHz from center frequency.

• Out-of channel noise. This parameter indicates how much a downstream data channel will interfere with adjacent video channels. Noise shouldn’t exceed –12 dBmV for any 6-MHz bandwidth outside the downstream channel.

• Output frequency. Measurements of the output frequency verify whether a CMTS transmits on the proper downstream channel frequency. The center-channel frequency must be within 630 kHz of the nominal value for that channel.

• BER (bit error rate). BER measurements must be made at a downstream data rate of 38 Mbps after a CMTS has applied FEC (forward error correction) to the downstream signal. BER should measure less than or equal to 10^–8 when the CMTS operates with a carrier to noise ratio (E_s/N_o) of 23.5 dB or greater.

Figure 1. A cable modem termination system transmits and receives data from hundreds of cable modems over a hybrid fiber coax cable network.

An MSO reserves one or more 6-MHz channels containing downstream data. Each channel can serve up to 8000 cable modems. A downstream data channel resides between two analog or digital TV channels (Fig. 2).

The center frequency of a downstream data channel can range from 93 MHz to 855 MHz. That’s 127 possible channel frequencies. With two possible modulation types—64 QAM and 256 QAM—there are 254 possible combinations of modulation and channel frequencies.

As part of physical-layer testing, engineers often test a CMTS transmitter at the high and low ends of the frequency range plus the center of the range. But some engineers test at more channel frequencies to cover the frequencies most often used by MSOs while still testing in a reasonable amount of time.

In the upstream direction, a CMTS must receive transmissions that use either QPSK or 16 QAM modulation over a frequency range of 5 MHz to 42 MHz (Fig. 2). Unlike downstream transmissions, which use a fixed 6-MHz bandwidth, upstream channel bandwidth can be any of five values from 200 kHz to 3.2 MHz, depending on the cable modem’s upstream data rate.

As “Range and Register” explains, a CMTS must tell a cable modem how much power to use when transmitting an upstream signal. Power adjustments ensure the received signal will be within a range that won’t overload the CMTS or other network components. If a signal entering an amplifier, an electrical-to-optical converter, or an optical-to-electrical converter is too strong, the device will clip the outgoing signal and distort it.

Figure 2. A downstream channel occupies 6 MHz and may be adjacent to TV channels. An upstream channel resides below TV channel frequencies and has a bandwidth from 200 kHz to 3.2 MHz, depending on the data rate. (Courtesy of RiverDelta Networks.)

After completing downstream and upstream physical-layer tests, engineers move to the MAC layer. Here, engineers measure parameters such as the effectiveness of FEC on an upstream transmission. A cable modem will send from 1 to 10 FEC bytes for every 16 bytes to 255 bytes sent to a CMTS; the FEC bytes correct for data errors in the upstream signal. Although the DOCSIS standards don’t require engineers to add interference to an upstream test signal, most manufacturers will add impairments such as bursts of wideband noise and narrowband carriers to the upstream channel so they can test the effect that FEC has on data throughput.

Engineers also measure the bit jitter that a CMTS produces, looking for a maximum deviation of 500 ns on a 10.24-MHz sync signal that’s encoded into the data. The timing of the sync signal ensures that upstream data from two or more cable modems won’t collide on the cable, which would destroy the data.

In addition to running prequalification tests, engineers also must run performance tests to ensure that a CMTS will communicate with the dozens of cable modem models available today. For performance testing, CMTS engineers test their designs using as many cable modems as possible under simulated real-world conditions in the lab. The performance tests include measuring data throughput, packet loss, and packet latency.

Test Equipment
To perform DOCSIS and performance tests, engineers need a variety of test equipment. Because DOCSIS testing involves many RF measurements, CMTS manufacturers typically use a vector signal analyzer such as the 89441 from Agilent Technologies (Santa Clara, CA). The analyzer measures the RF characteristics of the upstream and downstream signals. With those measurements, engineers verify channel bandwidth, amplitude, out-of-channel noise, and channel frequency of a downstream channel. The analyzer also can demodulate the channel’s signals and produce constellation diagrams of the 64 QAM and 256 QAM downstream signals or the 16 QAM and QPSK upstream signals. From those measured constellations, the instrument calculates EVM.

Each of perhaps more than two dozen downstream transmitters and more than 100 upstream receivers in a CMTS must transmit data to and from hundreds of cable modems. To avoid upstream data collisions, a CMTS uses TDMA (time division multiple access) to give each cable modem a time slice in which to send data. Therefore, each cable modem sends data in bursts, then goes quiet until the next scheduled time slice.

The CMTS must detect the beginning and end of each burst. An oscilloscope set to trigger on the length of a burst verifies that the CMTS receives data sent with the proper burst length.

To generate data traffic for both downstream and upstream physical-layer tests, the MAC-layer tests, and the performance tests, CMTS manufacturers need equipment that can simulate the aggregate traffic of hundreds of cable modems sending data to and receiving data from the Internet. Traffic generators such as those from Ixia Communications (Calabasas, CA) and from NetCom Systems (Calabasas, CA) generate IP packets for the CMTS to send and receive.

Traffic generators do more than transmit data packets. They also perform MAC-layer measurements for data throughput, packet loss, and packet latency. DOCSIS doesn’t require these tests for CMTS systems or for cable modems, but the measurements are essential for performance tests.

Testing to DOCSIS requirements is time consuming. The possible combinations of tests on RF channels, power levels, modulations, forward error correction, and others are nearly limitless. Although they can’t test for every possible combination, engineers still need to make hundreds of measurements. CMTS manufacturers often automate their DOCSIS tests by building or buying an automated test station. Automated DOCSIS testers are available from system integrators Agilent Technologies, DAQtron (Roswell, GA), and Symtx (Austin, TX).

Making an HFC Network
To run performance tests, CMTS engineers need to simulate the HFC (hybrid fiber coax) network that makes up a cable plant. Figure 3 diagrams a typical in-house HFC network that engineers use for interoperability and stress testing. The network includes all the components that make up a network from the CMTS to subscribers’ homes.

Figure 3. CMTS interoperability testing often incorporates an in-house cable network with a data generator to simulate Internet downstream traffic and subscribers’ upstream traffic.

The network consists of tens of kilometers of fiber cable in spools. Typically, engineers at a manufacturer’s CMTS test lab will use several rolls of fiber and coax to create several cable lengths. It’s important that the engineers not just test at the longest possible cable lengths. Because the cable modems send data upstream in time slices, the CMTS must compensate for transmission delays, and it must communicate with modems at many different distances.

A CMTS manufacturer’s performance test lab also needs cable modems—hundreds of them—so engineers can verify the interoperability of a CMTS and perform stress tests under heavy data loads. CMTS manufacturers have racks full of cable modems lining the walls of their test labs.

The coax cable that simulates a trunk connects to the racks of modems on their RF sides. As many as four amplifiers may reside along one cable trunk. A single trunk can support about 500 modems. Additional rolls of coax attach to the amplifiers but don’t connect anywhere. These rolls simulate unterminated cable lines reserved for new subscribers.

Until this year, no equipment existed that could simulate an HFC cable network. CMTS makers now have an option of using an impairment emulator from Telecom Analysis Systems (TAS, Eatontown, NJ) to emulate a cable network. A cable-network emulator can generate several impairments, including intermodulation distortion, group delay, and amplitude tilt. Engineers that already have their own in-house HFC cable networks use the networks in conjunction with the impairment emulator. The emulator lets engineers test a CMTS under controlled, repeatable conditions.

Engineers can also use a noise-burst generator from NoiseCom (Paramus, NJ) to add impairments to data transmissions. The instrument simulates RFI that can interfere with upstream transmissions. Most interference to upstream transmissions comes from the home. Appliances, tools, hair dryers, and other products generate RFI.

As part of performance testing, engineers need to generate video signals on channels adjacent to a CMTS downstream channel. (No video signals reside in the same frequency range used for upstream transmissions.) A test facility may use a comb generator to generate carriers that simulate those of analog TV signals. Engineers verify that the adjacent video carriers don’t interfere with downstream transmissions and that downstream transmissions don’t degrade the TV signals.

Once a CMTS manufacturer sets up an in-house test lab, it must work to keep the lab up to date. In particular, engineers need to constantly expand and update their collection of cable modems. Each time a cable modem manufacturer introduces a new model or changes the firmware in an existing model, CMTS manufacturers must get the new modem and verify that their products communicate with it. MSOs assume that a CMTS will communicate reliably with any cable modem. If a CMTS won’t talk to a cable modem, engineers at an MSO will assume the problem lies with the CMTS and will call the CMTS manufacturer for help. T&MW

FOOTNOTE
1. Data over Cable Service Interface Specifications, Cable Modem Termination System—Network Side Interface Specification, document number SP-CMTS-NSII01-960702. CableLabs, Louisville, CO, 1996. You can download the DOCSIS documents at www.cablemodem.com   (click on the “Public Entry” button).

FOR FURTHER READING
SmartBits Guide to DOCSIS Testing, Application Note 26, NetCom Systems, Calabasas, CA, May 2000. www.netcomsystems.com/technology/appnotes/AppNote26_
DOCSIS.pdf.

Smith, Gregory, “Cable Modem Certification,” Communications System Design, November 1999. www.csdmag.com/main/1999/11/9911feat1.htm.

ACKNOWLEDGEMENTS
I’d like to thank the following people for reviewing this article: Tom Gallagher, director of the broadband test program at CableLabs; Paul Nikolich, vice president of technology and standards at Broadband Access Systems; Mike Pellegrini, product manager at Telecom Analysis Systems; and Nick Signore, director of technical support at RiverDelta Networks.

Range and Register A CMTS contains interface cards for connecting it to the Internet and to the rest of the cable network. (Courtesy of Broadband Access Systems.) When a subscriber connects a cable modem to the cable network, the modem goes through a process called “ranging and registering.” In effect, the CMTS detects a cable modem and adjusts (ranges) the modem’s upstream signal power until the CMTS receives power at the proper level. Then, the CMTS can register the cable modem by sending it an IP address.   Once a cable modem has completed the ranging process, its actual upstream transmit level will fall between 8 dBmV and 58 dBmV for QPSK modulation and 8 dBmV to 53 dBmV for 16 QAM—a 47-dB or 50-dB dynamic range. Typically, CMTS manufacturers design their products to operate from 18 dBmV to 48 dBmV, allowing a 10-dBmV safety zone. Some CMTSs are designed with a 5-dBmV safety margin and can operate from 13 dBmV to 53 dBmV. The minimum power level where the CMTS can reliably receive upstream data is called the target level, and CMTS manufacturers build a tolerance into that level, which is typically 3 dB. After settling the received power level to within the target level, the CMTS adjusts each cable modem’s output power once every 35 s.   A CMTS must broadcast several special support packet types to all the cable modems on the HFC network. The support packets provide timing information, upstream channel information (modulation type, frequency, and bandwidth) and schedule information telling each modem when it can send data.   A CMTS uses time-division multiplexing to send these packets to cable modems between downstream data transmissions. A cable modem must continue to see these support packets within specific maximum time intervals. Otherwise, the cable modem will initiate corrective action that will interrupt its ability to pass Internet data to its host PC.—Martin Rowe