Mitigating EMI

Wireless Interference in a Central Office Environment

David A. Case

Sometimes the real challenges on a wireless project occur after it is deployed in the field—beyond the controls of the lab environment.

Photo courtesy of CISCO SYSTEMS (Richfield, OH)

Sometimes problems occur after wireless equipment is already deployed on customer premises. In these cases, the manufacturer must not only solve the interference issue but must also address the customer's concerns. This can present a tough challenge because it is often not the wireless device acting up, but rather the customer's equipment that is not properly immune from the nearby transmitters, specifically portable transmitters.

This same situation transplanted into a central office environment could be a disaster. Shutting down a phone network is a real concern.

The Problem

The issue is whether wireless devices would disrupt the equipment in a central office environment. Most equipment installed in a central office environment is required to be tested for network equipment building system (NEBS) compliance. This article discusses an on-site evaluation of wireless devices that was conducted to assess their effect in the intended environment.

The carriers discovered some network interference issues when they deployed 802.11b wireless in their central offices. Before deployment, they hadn't performed due diligence or EMI studies. The carriers' senior management required that the engineers address the interference issues before they could continue with the wireless rollout. We worked with their engineering staffs to troubleshoot the situation. We then developed a generic way for the field technicians to predict and test for interference.

There are no actual in situ test procedures for this type of testing except ANSI C63.18, which addresses the medical environment. Therefore, in order to solve the issue, a repeatable test procedure had to be developed for use in testing the equipment in the central office environment.

In addition to assessing the equipment's ability to operate normally, testing was also conducted with the covers removed and cabinet doors opened.

Both of these conditions generated some very interesting results. From these observations, new testing methods were developed for testing and mitigating interference.

With the requirements for NEBS compliance, EMI issues in a central office environment should be of less concern than in other environments. However, that was not the case in this evaluation. With the cabinets open, the devices were no longer technically NEBS compliant. They had never been evaluated for that mode of operation. NEBS testing is all done at 3 m. In this test, the customer wanted to insert the antenna into the cabinet to see whether a problem existed.

Technical Discussion

To better understand the radio-frequency (RF) interference issue, it is important to know what FCC means by harmful interference. As defined in 47 CFR Part 2 and Part 15 Sections 2.1 and 15.3(m), harmful interference is an emission that endangers the functioning of a radionavigation service or seriously degrades, obstructs, or repeatedly interrupts a radiocommunications service.

A Part 15 device that is being interfered with, either by another Part 15 device or by any licensed radio service, cannot claim protection. Even though Part 15 says the device must accept interference, in the United States and Canada, there are no mandatory regulations requiring the testing for compliance to demonstrate this for Part 15 devices. The requirements to comply with specific susceptibility standards such as Telcordia Generic Requirement (GR-1089-CORE) are industry driven rather than regulatory driven.

Most equipment in a central office generally operates without error when a transmitter is several meters away. This is because most central office equipment is evaluated for susceptibility from transmitters operating in the far field. This testing is done as part of GR-1089-CORE, which tests devices for normal operation when exposed to E-field strengths of 8 or 10 V/m.

Although central office equipment will most likely not be affected by the wireless access point equipment, it could very well be affected by a portable transmitter operating in close proximity to the device. In this instance, the portable transmitter would be the mobile handheld device interfacing with the access point equipment. This effect is known as near-field coupling, the most common source of interference problems. Near field is defined as a spherical area with a radius that is one to two times the radio wavelength, written as

The higher the radio frequency, the shorter the wavelength and the smaller the near-field effect. An illustration of the difference in near-field range between two operating frequencies is shown in Figure 1.

Figure 1. An illustration of the difference in near-field range between the two operating frequencies, 5 GHz and 2.4 GHz.

Test Procedure

A procedure was developed for in situ ad hoc testing in a central office environment. GR-1089 does not require tests for the effects from near-field devices; its requirements address only far-field measurements, and the standard does not use complex waveforms such as the orthogonal frequency division multiplexing (OFDM) modulation scheme. We observed that the OC48 modules did not have a problem with an 80% modulated AM signal when measured at 6 in. at 5 V/m; however, they would lock up when at an OFDM signal greater than 3.5V/m at any distance.

Furthermore, GR-1089 is ineffective when the covers of the Telco equipment are removed during service. This was the biggest hindrance to solving the issue. When the cabinets were opened, the network would go down if the transmitters were within several inches. The carrier found this unacceptable. In addition, two specific pieces of older gear that had been tested to an earlier version of GR-1089 had issues with 2.4 GHz.

The carrier needed an on-site test method to verify  that interference will either not occur (GR-1089-CORE does not guarantee that no interference will occur) or a test that would at least determine which areas to restrict from the operation of portable wireless devices. This required us to develop a test in a real-life situation that would enable a field technician to verify there were no problems.

We had to address the interference caused by systems with cabinet doors open with part of the rack operating normally and part of it not operating normally (these scenarios are according to the Cisco best practice guide).

According to GR-1089-CORE emissions, if the Cisco radio meets the GR-1089 emissions standard in the 2.4 GHz or 5 GHz U-NII  bands, it is called "broken." The fundamental power of a Cisco transmitter exceeds GR-1089-CORE emissions limits by their very nature. A Cisco wireless device can operate at up to 4 W in parts of the bands.

Cisco has previously written an ad hoc test procedure for its medical guide for use in testing its WLAN in the medical environment. That procedure is based on the more generic ANSI C63.18 medical test standard. This Cisco document was used as the basis for developing the test procedure for this evaluation. The goal was to develop a procedure that customers could use in the field to determine whether any problem existed. It was designed to be a fairly quick procedure without being overly burdensome.

For this test, the radio mode of operation was to perform a file transfer using an access point. This simulated a typical real-world scenario.

For this particular evaluation, the procedure used a total of 16 access points around the equipment under test (EUT). The number of access points can vary. The measurement started with the radio at about 1 in. from the EUT at the first test point, and the final measurement was done at 1 m.  A sliding scale was used at 1 in., 3 in., 6 in., 1 ft, 1/2 m, etc. Other distances can be used. Testing was done at one test angle first, and then upon completion of all test distances, the test was repeated again at the next test point. It is critical to observe and record any anomalies that occur during the testing.

Interference Issues

It was determined that a 2.4 GHz WLAN was more disruptive to some of the systems than a 5 GHz U-NII  WLAN. There were several reasons for this result.

First, some of the central office equipment had internal clocks that were running at or close to 2.4 GHz, which made them more susceptible to interference when the boxes were opened up. Second, the 2.4 GHz radios generally operated at higher power than the 5 GHz U-NII  radios that were evaluated. Part of this was due to the threshold of pain for these systems, which we determined was when the field strength exceeded 3.5 V/m. Hence, the 2.4 GHz systems were more disruptive when in close proximity (less than 8 in. ) than the 5 GHz U-NII  systems were at 3 in. It is important to note that we were not simply reinventing the wheel.

We had to provide not only a prediction model, but also a test methodology. As requested by the customer's senior management, both the prediction model and the test method had to be easily understood. Therefore, the approach was simple and very effective. Until this point, no one had documented these procedures in a best practice guide for deploying wireless LAN in a central office environment.  In each case, all wireless projects are going forward, and calls reporting interference have stopped. That was the goal.

Solution

It was determined in this particular case that deploying a 5 GHz U-NII  radio rather than a 2.4 GHz radio reduced the risk of RF interference. RF interference not only correlated to proximity, but also correlated to the level of transmitter RF power and frequency. The higher the RF power, the higher the probability of RF interference at the same distance. Therefore, reducing the RF power did help to mitigate RF interference.

Further, it was determined that a minimum separation distance was required between a wireless handheld device and the central office equipment being tested. To determine a uniform separation distance, it was necessary to determine the precise field-strength level that caused interference to the EUT.

In reviewing the testing and simulation data and with respect to the anomalies observed on some legacy equipment, it was determined that the system operated without error when the E-field was equal to or less than 3.5 V/m for the 5 GHz radio transmitter. The field strength of the 5 GHz radio emissions at 5 in. with a power output of 8 dBm into a 2.1-dBi dipole antenna was found to be 3.25 V/m (see Table I and Figure 2). Issues arose when the 5 GHz device was either operated at a higher power or at a closer proximity than 5 in. (where the E-field would exceed 3.5 V/m).

Figure 2. Near fields from 5300 MHz dipole. Far-field behavior begins after 3–4 in.

Going Forward

The use of wireless devices in a central office environment is challenging. It is even more so if the requirement is that no interference can occur when equipment is open for service or testing.

The wireless industry needs to develop a standardized test procedure for testing wireless devices within central office environments. The development of an industry test standard for near-field measurements is currently being undertaken by the industry. As part of this effort, the procedure used for this testing has been forwarded to the relevant standards committee.

In addition, central office operators need to consider developing operational procedures or best practices for wireless use, specifically for handheld radio devices. These practices should include either exclusion zones or minimum separation distances between wireless devices and central office equipment.

Acknowledgment

The graphs and near-field-emission study data were provided by Steve Saliga and Fred Anderson of Cisco Systems Wireless Business Unit.

David A. Case, NCE, NCT, is senior reg-ulatory engineer, corporate compliance EMC standards and operations, for Cisco Systems Inc. (Richfield, OH). He can be reached at davecase@cisco.com.