Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
7 C H A P T E R 2 Wireless communication, most commonly encountered through the use of either wireless fidelity (WiFi) or the cell phone, has become a routine tool for daily life. However, wireless com- munication is subject to radio frequency (RF) interference. The increasing use and importance of wireless, not only as a passenger amenity but as an integral part of airport operations, makes its reliability and performance critical for modern airport operations. Security is critical too because wireless has become a potential attack vector, a way to disrupt an airportâs operation. Thus, a robust secure network that is well-managed and upgraded in accordance with increased demands and applications is a major concern for airport managers. As part of the research for this Guidebook, team members collected information from airports regarding their experience with WiFi, in particular problems with interference, capacity, and performance. The team was also interested in what solutions had been tried and whether they were successful. Through a distributed survey completed by 18 airports as well as site visits to 9 airports, the researchers gathered background information that helped inform this chapter. This chapter describes the technical aspects of RF interference, the associated issues, and the potential solutions. While the primer (Appendix A) is written with airport management in mind, it is the information technology network engineer (or contractor carrying out this responsibility) who is the targeted audience for this chapter. Included are processes, techniques, procedures, and applicable tools that can be used to isolate the cause of interference and define solutions to mitigate the problem. Readers will be encouraged to ascertain when interference may be due to causes other than RF emanationsâcauses such as network congestion, equipment interoperability problems, or simply poor coverage. This chapter also provides recommenda- tions on alternate methods and resources that can be accessed when the problem exceeds the engineerâs usual efforts. For smaller airports, which typically do not employ full-time systems engineers, there is a separate chapter, Chapter 6 WiFi at Small and General Aviation Airports. The most appropriate solutions for any given airport will depend upon the airportâs goals for WiFi service and the amount of funding available to correct problems and expand capacity. WiFi at Airports WiFi service level requirements have rapidly gone from best-effort service being broadly accepted to the current expectation by many travelers that wireless networks have reliability and performance close to that of a wired network. In the early days of WiFi, passengers were pleased to have any wireless service at all and they expected to pay for it. Today, passengers expect con- nectivity everywhere, with service levels similar to a wired connection or at least equal to the wireless service they enjoy in their homes. As airports and their tenants integrate wireless con- nectivity into their operations, even higher service levels are necessary and justified. WiFi Service at Airports and the Problem of Interference
8 A Guidebook for Mitigating Disruptive WiFi Interference at Airports Service expectations of WiFi networks vary widely among both airport network managers and travelers. Some airport managers and travelers have come to view WiFi as part of the necessary infrastructure. They expect WiFi to work well, just like they expect the lighting to be good and restrooms to be clean and functioning. On the other hand, some airport managers place a lower importance on their airportâs wireless network capacity than they do other management issues demanding attention. Travelers with high expectations for WiFi service will view poorly an airport that does not meet their expectations for seamless communications service. Further, as there are limited opportunities for airport managers and travelers to negotiate and align their expectations, air- ports that do not provide high quality WiFi may find that a percentage of travelers may choose other airports for departing or connecting flights. Business travelers in particular need reliable Internet connectivity to do work while waiting for flights, especially when flights are delayed or canceled, disrupting business schedules. At some studied airports, the WiFi network is still viewed primarily as a passenger amenity and service expectations are low. In contrast, others view their network as part of their core infra- structure. These airports tend to take the travelerâs experience very seriously and recognize that the quality of a passengerâs experience with the network is part of their total airport experience. Network Management Arrangements Not surprisingly, there are a variety of network management arrangements. Some airports manage their own WiFi and work directly with cellular network providers for cellular coverage in the terminal. Others contract out both the WiFi and cellular network management, delegating the relationship with cellular providers to their chosen wireless network management company. Once the company is selected, some airports view their responsibility for network management as over, until the next contract negotiation. Other airports have found they must increasingly become more involved in network management and have a close working relationship with their chosen vendor. At several of the largest airports visited, however, airport managers noted they did not have the level of expertise necessary to adequately oversee the work of their network management vendor. These airports have contracted with a second vendor that has in-depth RF and network management expertise to provide measurements and independent input on network operation and the service level provided. Many airports run parallel services. The free service offers limited data speeds and often requires users to listen to commercials before being allowed to access the Internet. In parallel, paid service is available for a fee, paid either per instance or by subscription. This essentially is attempting to model WiFi after the cellular network. Travelers have a choice. They can access the Internet through their cellular devices and share that connection with their other devices easily. Table 1 summarizes key aspects of WiFi at nine airports that participated in the case study analysis. Causes of WiFi Interference and Disruption Operational Definition What is considered RF interference? This is not as straightforward as might be imagined. For this study, degradation in performance or disruption of communication was considered interference. A disruption of communication is easy to understand. Users cannot connect to the network, but they should be able to. Something is blocking or disrupting the communication. A degradation of performance is more involved. If packets are lost due to another transmitter,
Table 1. At-a-glance WiFi summary of case study airports. Topic ABIA ACT BOS BWI DFW GRK JAN LAX SEATAC Responsible authority City of Austin Information Technology Department City ofWaco, TX Massachusetts Port Authority Maryland Aviation Admin. Dallas/Fort Worth Airportâs Department of Information Technology City of Killeen, TX City of Jackson,MS Los Angeles World Airports Port of Seattle (a public corporation under King County) WiFi network management vendor Boingo (and subsidiary Concourse Comm.) In house AWG (now part of Boingo) Boingo AT&T In house Wandering WiFi AWG (now part of Boingo) In house (Boingo provides advertising only) Flights/day 483 96 882 734 1854 31 137 1674 868 Passengers/year 10,017,158 61,401 30,236,088 22,501,353 60,436,266 174,000 1,200,000 66,667,619 34,824,281 Access point vendor and type CISCO CleanAir CISCO 1 access point CISCO Aruba CISCO 1,000 access points Various 8 CISCO Aironet; 1100 access points CISCO CISCO Bandwidth to Internet 200 Mbps Not provided Not provided Not provided Approaching 1,000Mbps Not provided Not provided Not provided Not provided Operatingmetric 95% coverage 95% of the time Nometrics developed Customer complaints that reach MassPort and complaints to help desk not tracked by MassPort No routine reporting or testing of network performance or capacity. Customer complaints are monitored via social media and addressed. User experience is the central metric used. They translate this tomean 8 10 Mbps for every user of the free WiFi. Customer complaints, however service is judged against a low quality of service expectation because the service is free. Customer complaints judged against a low expectation for quality of service because the service is free. Customer complaints Coverage/Band width/Protocol: Minimum 2 access points at gates (only 6 in whole concourse). Putting in 24 access points in baggage claim area and expanding concessions signiï¬cantly. 5 Mbps up and down Note: Mbps= megabits per second.
10 A Guidebook for Mitigating Disruptive WiFi Interference at Airports that is interference. If the communication is slower than it should be when another device is transmitting, that also is considered interference. There is a difference between interference and interference that causes problems. Some airports run more of their operational functions over the network and they need those functions to operate reliably. Others compete with other airports for passengers and the quality of the travelerâs experi- ence is extremely important to them. There are many factors that affect an airport authorityâs expec- tations for connectivity and whether they view the degree of interference as problematic or not. Intentional Interference The intentional creation of interference has already become reasonably common. It is a grow- ing security risk that must be considered. In the research for this project, one airport reported that occasionally people come to the airport and set up a rogue hotspot using the airportâs service set identifier (SSID) in an attempt to get peopleâs login and password information. Rogue hotspots can operate at higher output power than is permitted in the 2.4 GHz unlicensed spectrum and, when doing so, cause interference. Airport managers reported that their network operator had techniques that allowed them to use the network to suppress these fraud hotspots when detected. Presumably, since networks regularly turn their power down to avoid interference, they can also turn their power up to create interference for a fraudulent hotspot. That this airportâs network manager had good motives still leaves the fact that people with bad motives can create interference. As wireless becomes increasingly integrated into an airportâs operation, it must also become part of the security planning at the airport. This is particularly true if security cameras and access control devices are connected wirelessly, because a cyber-attack could suppress the wireless con- nection and therefore circumvent the security system. When thinking about a security issue of this type, it is typical to think about detection, preven- tion, and mitigation. If intentional interference is created, how will it be detected and who will be notified? What can be done to prevent intentional interference? If efforts to prevent interference fail, what can be done to mitigate its impact? Finding and implementing answers to these ques- tions becomes increasingly important as the use of wireless grows and evolves, becoming more deeply integrated into the infrastructure. 2.4 GHz Band Congestion One of the most striking findings from this study was the congestion in the 2.4 GHz band. Although there are many more channels and much more spectrum in the 5 GHz band, by far the majority of the traffic is packed into the 2.4 GHz band. To make matters even worse, traffic was repeatedly congested into a signal channel in the band. For both the WiFi and cellular networks, the ability to use new frequencies in the 5 GHz bands opens up the opportunity to spread users out, separating them in frequency so that they can receive better service and avoid interfering with each other. However, that opportunity is appearing to wither in the presence of market forces that congregate devices into the 2.4 GHz band. It appears as if market dynamics are equipping the majority of devices to only operate in the 2.4 GHz band. A natural consequence of this is that the 2.4 GHz band is becoming increas- ingly crowded in many locations where multiple users congregate. Further, dual-band devices generally default to this crowded band, even though they are capable of operating in the much less crowded 5.8 GHz band. The use of WiFi and Bluetooth, quickly being joined by ZigBee, on the 2.4 GHz ISM band, has become the dominant choice for product designers who want to include a wireless interface
WiFi Service at Airports and the Problem of Interference 11 in their product. In a recent technology trends article published in Electronic Design, âBluetooth and Wi-Fi Rule the Airwaves,â Louis E. Frenzel stated: With dozens of short-range wireless standards to choose from, engineers still defer to Bluetooth and Wi-Fi. After more than 15 years, they continue to grow, improve, and provide new features and benefits. These amazingly useful technologies both use the 2.4-GHz industrial-scientific-medical (ISM) unlicensed spectrum.1 Although the 5.8 GHz band is available and offers more bandwidth and higher data rates, most product designers only equip their products to operate in the 2.4 GHz band. As a result, interfer- ence is a growing problem in the band and devices equipped to operate in the 2.4 GHz band do not have the option of moving to a less congested band when they encounter congested conditions. The data in Table 2 were gathered from the WiFi Alliance database of WiFi certified devices. Devices qualified for IEEE 802.11b are assumed to support only the 2.4 GHz band, while devices that are qualified for both IEEE 802.11a and 802.11b support both the 2.4 GHz and 5.8 GHz bands. All Certified Wifi deviCes Total certified WiFi devices: 15,748 Total devices with 802.11a capability: 5,471 Percent of all devices with 802.11a capability: 35% As Figure 6 and Figure 7 demonstrate, 77% of network access points are operating in the 2.4 GHz band and 80% of the traffic is using this band. However, the 2.4 GHz band only has 3 non-overlapping WiFi channels versus 25 channels in the 5 GHz band. That means that 80% of the traffic and 77% of the network access points are crowded into 10.7% of the available chan- nels, leaving the remaining 89.3% of the channels to carry only 23% of the traffic. The crowding is compounded by WiFi traffic being crowded into only 3 bands, most domi- nantly WiFi channel 11, which carried 44% of the traffic in this sample (Figure 8). When channels become congested and suffer interference, their data rates slow down and the error rates go up, resulting in data retransmissions. Although packet retransmissions occur routinely in WiFi communications, the common metric is that retransmissions greater than 1% are a symptom of interference in the channel. Overcrowding in the 2.4 GHz band results in lower data rates and a higher level of transmission errors, which require a large number of packet retransmissions. However, as Figure 9 makes clear, high rates of packet retransmission are com- mon at airports and represent data packet captures of over 1500 channels at 162 locations at Year Dual BandDevices Single Band Devices % Dual Band 2013 1405 2826 50% 2012 1425 3445 41% 2011 1016 2885 35% 2010 582 1975 29% 2009 388 1197 32% 2008 249 818 30% 2007 218 724 30% 2006 115 501 23% Table 2. Single vs. dual-band WiFi devicesâall certified devices. 1 Louis E. Frenzel, âBluetooth and Wi-Fi Rule the Airwaves,â Electronic Design, July 11, 2013. Figure 7. Traffic distribution by band. Figure 6. Access point distribution by band.
12 A Guidebook for Mitigating Disruptive WiFi Interference at Airports WiFi Traffic Distribuon by Channel All 5GHz traffic 20% Channel 11 44% Channels 7 10 5% Channel 6 9% Channels 2 5 9% Channel 1 13% Figure 8. Traffic distribution among WiFi channels. 28 airports; retransmission rates of over 10% and as high as 70% are relatively common on airport WiFi networks. The situation becomes even clearer when the retransmission rates for channels in the crowded 2.4 GHz band, which were only able to achieve data rates under 6 MBs (Figure 10), are com- pared to those channels in the less crowded 5 GHz band that were able to achieve data rates of ⥠24 MBs. When channels become congested and suffer interference, their data rates slow down and the error rates go up. Figure 10 suggests that slow channels in the 2.4 GHz band are commonly suffering interference. In contrast, channels in the less crowded 5 GHz band achieve much higher data rates with a far lower error rate (Figure 11). Figure 9. Percent of packet retransmission as a function of access point (AP) load at 28 airports. Packet Retransmission as a Function of AP Loading Packets Per AP Pa ck et R et ra ns m is si on s ( % )
WiFi Service at Airports and the Problem of Interference 13 Figure 10. Percent of packet retransmission as a function of access point (AP) load at 28 airports for channels in the 2.4 GHz band with < 6 MBs data rates. Packet Retransmission as a Function of AP Loading 2.4 GHz - < 6MBs Packets Per AP Pa ck et R et ra ns m is si on s ( % ) Packet Retransmission as a Function of AP Loading 5 GHz - ⥠24MBs Packets Per AP Pa ck et R et ra ns m is si on s ( % ) Figure 11. Percent of packet retransmission as a function of access point (AP) load at 28 airports for channels in the 5 GHz band with >â 24 MBs data rates.
14 A Guidebook for Mitigating Disruptive WiFi Interference at Airports A result of the 2.4 GHz band crowding is low data rates, reported in Figure 12, and a high level of transmission errors requiring packet retransmission, reported in Figure 13. The realized data rates are very slow compared to the specified maximum rates for WiFi, and those in the 2.4 GHz band are approximately half of those experienced in the 5 GHz band. Transmission errors are also much more common in the 2.4 GHz band, as seen in Figure 13. It is noteworthy that WiFi channel 6 has an unusually high percentage of error when com- pared to WiFi channels 1 and 11, which carry a higher percentage of the traffic. It is believed this is because WiFi channel 6 has overlapping WiFi channels to either side, while channels 1 and 11 only have that situation to one side. This doubles the potential for adjacent channel interference, which appears to be a real problem for channel 6. Average Data Rates by Channel (Mbits/s) 7. 86 3 8. 48 8 2. 72 3 6. 45 5 7. 43 5 5. 98 0 3. 86 6 3. 80 7 5. 43 2 3. 43 4 9. 27 9 14 .6 82 13 .8 40 17 .3 76 14 .4 12 14 .4 33 22 .4 39 18 .8 92 16 .8 55 1 9. 32 4 23 .4 97 10 .8 88 10 .3 05 17 .0 67 12 .3 17 13 .9 18 12 .5 60 14 .9 79 9. 07 1 1 2 3 4 5 6 7 8 9 10 11 36 40 44 48 52 56 60 64 10 0 10 4 10 8 11 2 13 6 14 9 15 3 15 7 16 1 16 5 WiFi Channel Figure 12. WiFi data rates by channel observed at 28 airports. Figure 13. Packet retransmission rates observed at 28 airports. Percent Retries by Channel 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 20% (r et ry p ac ke ts /t ot al p ac ke ts ) Channel 1 Channels 2-5 Channel 6 Channels 7-10 Channel 11 All 5 GHz traffic
WiFi Service at Airports and the Problem of Interference 15 An examination of WiFi use at airports shows that the trend toward band congestion is pronounced. Table 3 gives a sampling of the band usage at 28 airports. Data were gathered at 173 airport locations, mostly at the gates. These data represent the distribution of access points, and show how the airport network makes itself available on a band basis. A quick scan shows congestion lies in the 2.4 GHz band. A significant variation between airports is also seen. Figure 14 shows the distribution of band usage as measured at 26 U.S. airports and 2 European airports, including 162 gates and 11 additional locations, such as baggage claim. Although this is just a snapshot in time, the results illustrate that 73% of usage is crowded into the 2.4 GHz band. This is indicative of the results experienced throughout the study. Airport networks have distributed their access points in ways that generally follow the distri- bution of consumer devices. However, this has created a vicious cycle that contributes to over- crowding into the 2.4 GHz band. If the congestion in the 2.4 GHz band is to be reduced, then Airport 2.4 GHzChannels 1 14 Amsterdam Airport Schiphol (AMS) 100.00% Austin Bergstrom International Airport (AUS) 54.30% Baltimore Washington International Airport (BWI) 71.29% Birmingham Shuttlesworth International Airport (BHM) 73.68% Charlotte Douglas International Airport (CLT) 88.00% ChicagoMidway International Airport (MDW) 61.62% Chicago OâHare International Airport (ORD) 63.75% Copenhagen Airport (CPH) 100.00% Dallas/Fort Worth International Airport (DFW) 57.61% Dallas Love Field (DAL) 66.90% Denver International Airport (DEN) 76.20% Hartsï¬eld Jackson Atlanta International Airport (ATL) 96.72% John Wayne Airport, Orange County (SNA) 71.86% Killeen Fort Hood Regional Airport (GRK) 100.00% Logan International Airport (BOS) 75.75% Long Beach Airport (LGB) 68.70% Los Angeles International Airport (LAX) 58.41% McCarran International Airport (LAS) 66.37% Minneapolis St. Paul International Airport (MSP) 100.00% Nashville International Airport (BNA) 81.96% Newark Liberty International Airport (EWR) 87.50% Oakland International Airport (OAK) 64.09% Philadelphia International Airport (PHL) 96.64% Reagan National Airport (DCA) 81.34% Seattle Tacoma International Airport (SEA) 75.39% Tampa International Airport (TPA) 64.56% Waco Regional Airport (ACT) 100.00% William P. Hobby Airport (HOU) 90.57% Table 3. Sample of use of the 2.4 GHz band at access points at 28 airports.
16 A Guidebook for Mitigating Disruptive WiFi Interference at Airports both more opportunities to connect in other bands must become available and an increasing number of network users must be incentivized to use those opportunities. Adjacent Channel Interference Radio frequency devices do not have perfect frequency boundaries. While they are designed to put as much of their energy as possible into the channel they are using, there is an influence on or from devices operating on nearby channels. For a WiFi transmitter, some of its energy will spill over into nearby channels, which will add noise to those channels and reduce their ability to com- municate with the device they are intending to connect to. When receiving the signal, filtering and front-end RF circuitry allow some energy from an adjacent channel transmission to come in and influence the WiFi receiver. Figure 15 shows the way energy from an adjacent channel transmis- sion may affect a WiFi receiver and is often the dominant effect on a WiFi deviceâs performance. When two WiFi devices are close and have good signal strength between them, there is an operating margin and the impact of adjacent channel transmissions will be much less, perhaps negligible. The worst problem occurs when a WiFi device is trying to communicate over distance, resulting in a weak intended signal, but the adjacent channel transmission is close, resulting in the energy spill-over being much higher, at times high enough to totally prevent communications. As can be seen in Figure 15, there are two contributors to adjacent channel interference. Both must be improved for WiFi devices to operate more reliably. WiFi transmitters must reduce the amount of energy they spill over into other channels. However, that alone will not be enough. It is also necessary that WiFi receivers have better filters which give them improved resistance to adjacent channel transmissions. Other Sources of Interference An emerging source of interference is largely being created by the need to support two net- works in an increasing number of frequency bands. As a result, antennas for the WiFi and cel- lular networks are often placed close to each other, and at times they share the same antenna Figure 14. Distribution of band usage at 162 airport gates.
WiFi Service at Airports and the Problem of Interference 17 in a shared distributed antenna system. Strong RF signals can create intermodulation products and a variety of related problems. These issues are well-known to the military, where many RF sources are often crowded close together on aircraft and ships. The same problems are starting to emerge in airport systems. Another class of interference problem is created by technology changes. In some cases, inno- vations made to reduce interference actually end up causing more interference. An example of technology changes causing interference is IEEE 802.11b. Newer versions of the IEEE 802.11 standard have moved away from direct sequence spread spectrum modulation to orthogonal frequency-division multiplexing (OFDM) modulation. However, support for the older IEEE 802.11b was necessary for backward compatibility. In this research, it was not uncommon to still find IEEE 802.11b devices operating, but when they do they often become sources of interference because they are mismatched with how most WiFi devices currently operate. Interference Mitigation Techniques Mitigation measures need to be selected based on what is affordable and at the same time most likely to resolve the interference to the particular type(s) of WiFi problems at that airport. Some airports will find these techniques completely reasonable and readily adopt them, while other airports may view them as challenging or unaffordable. Multiple-input, Multiple-output Multiple-input, multiple-output (MIMO) is a promising technology that can contribute to the airport environment. Multiple-input, multiple-output is a wireless technology that uses multiple transmitters and receivers to transfer more data at the same time, taking advantage of multipath propagation, in which transmitted signals bounce off walls, ceilings, and other objects and reach the receiving antennas at different angles and times. Multiple antennas work together to enable them to combine data streams arriving from different paths and at different times to increase receiver signal-capturing power, which increases data throughput and mitigates poten- tial interference from reflected signals. Multiple-input, multiple-output uses multiple antennas at both the transmitter and receiver to transfer more data at the same time to improve communication performance. It offers significant increases in data throughput and link range without additional bandwidth or increased transmit Figure 15. The effect of adjacent channel interference (ACI).2 2 Texas Instruments White Paper, âThe Effects of Adjacent Channel Rejection and Adjacent Channel Interference on 802.11 WLAN Performance,â SPLY005âNovember 2003.
18 A Guidebook for Mitigating Disruptive WiFi Interference at Airports power by spreading the same total transmit power over the antennas to achieve an array gain that improves the spectral efficiency (more bits per second per hertz of bandwidth) and the link reli- ability. MIMO is part of modern wireless communication standards such as IEEE 802.11n (WiFi). In IEEE 802.11n, MIMO is used to achieve maximum speeds by allowing devices to connect using multiple data streams spread in frequency. However, MIMO is also an interference miti- gation technique. If a device can connect on multiple frequencies and one of those frequency channels is being interfered with, even though the device may lose some connection speed, it can still communicate with the network on the frequencies that are not receiving interference. Multiple-input, multiple-output is now heavily integrated into the WiFi standards. For close distances with good signal conditions, multiple streams of data can be sent simultaneously to multiply the data rate. At far distances or under poor signal conditions, MIMO can be used as a type of antenna diversity, allowing a unit to pick the best signal quality from several options. Alternatively, advanced signal analysis techniques can be used to improve the total signal quality using multiple signal sources. A variety of antenna placements, using antennas with different beam width, as shown in Fig- ure 16, provide the tools for optimal area coverage and capacity provision. Antennas with direc- tionality can be used overhead to provide uniform signal illumination over the widest possible area. When overhead installation is not feasible, side-mounted antenna, usually with a slight downward tilt, can provide the needed coverage. Even-floor or under-floor placement is the right solution for some situations. In some situ- ations, it is even feasible to place an antenna under concrete and have it beam up through the concrete, as shown in Figure 17 and Figure 18. For any implemented WiFi solution, the airport authority should consider having the installed system tested to verify it meets performance and design specifications. In addition, periodic or on-going network performance testing is also recommended to monitor system performance under actual user conditions, which allows continual tuning of network performance and early warning of developing problems. 3 Lukaszewski, Chuck, âUltra-High Density WLAN Design & Deployment,â Wireless LAN Professionals Summit 2014, pre- sentation slide 9. 153 44 157 64 44 52 36 48 Figure 16. Optimal coverage can be achieved through the use of a variety of antenna placements with appropriate antenna beamwidths.3
WiFi Service at Airports and the Problem of Interference 19 Other WiFi and Network Problems Interoperability Issues Interoperability issues have also been sources of problems. Vendors implement the same requirements in different ways. Some systems have been reported to work well with one model of access points but prove to be problematic with another. At times, a vendorâs equipment has been known to cause false triggers, causing the network automation to malfunction. Another problem is when equipment manufacturers design equipment to use within a par- ticular network architecture. If the network designer uses that equipment in a different design architecture, then there may be interoperability issues. Equipment and functions that may work very well in one context may become more of a problem than a solution when operating in a network design that its manufacturer did not anticipate. Uncoordinated WiFi Use A serious problem for airport networks is the independent and uncoordinated use of WiFi. There are several categories of uncoordinated WiFi use at airports. From a management 4 Ibid., slide 10. 5 Ibid., slide 11. Figure 17. Floor mounting, under seating, is a good solution in some circumstances.4 Figure 18. Through-concrete placement can even work in some situations.5
20 A Guidebook for Mitigating Disruptive WiFi Interference at Airports viewpoint, there are those WiFi users that, with agreement between the airport and the users, potentially could be coordinated but currently are not coordinated. A second category of unco- ordinated use is those situations where getting an agreement between the airport and the user is unlikely to be feasible because the user is transient. This category contains travelers and contrac- tors who bring their own equipment to the airport but are only there for a short time. With WiFi users who are permanently at the airport or who regularly come to the airportâ tenants, airlines, and some contractorsâit is possible to come to a shared plan for coordinated use of WiFi. At some airports the majority of the WiFi users in this category participate in a single network managed by a neutral provider. To be effective, this approach must meet the cost and performance goals of all participants. If the neutrally hosted network costs substantially more than other options or fails to meet the performance objectives of some participants, then the airport is likely to pursue any of an increasingly wide range of alternatives. Frequent travelers are also a group that can potentially be attracted to a neutrally hosted net- work. The objective of HotSpot 2.0 is to lower the burden of accessing a network by providing an automatic login and a plan that works at multiple locations. Users can pay a subscription fee and their devices will automatically authenticate and log onto a participating hotspot when they are in range. If the vision for HotSpot 2.0 becomes a reality, frequent travelers will be able to pay a monthly subscription and their devices will automatically connect to networks at a wide number of airports and other locations. It should be observed that the coordination mechanism is very different in these cases. With a permanent vendor, there will almost certainly be discussions between representatives of the airport and the tenant or airline and some signed agreement. With frequent travelers, the HotSpot 2.0 provider will market the service to them and success will depend on how attractive the offering is. The last category of uncoordinated WiFi use involves those users who come to the airport infrequently and bring their own WiFi direct applications. This presents two different situations. With infrequent visitors, there isnât the opportunity to interest them in a subscription service unless it becomes very widely used by the general population. These users come with their WiFi devices and will use them. Many of these have mobile hotspots of various kinds and are using the cellular network for Internet access. Locally their device, which is often a smartphone but may be a dedicated hotspot device, will provide a local WiFi signal and relay the traffic to the Internet through the cellular network. This kind of service is a competitor to HotSpot 2.0 and users are likely to compare cost, performance, and convenience when deciding which service to use. Then there are an increasing number of device-to-device services that use WiFi for connectiv- ity. In these applications, one device is sending data to another device and really is only using WiFi because it is inexpensive and easily integrated into the products. Many of these devices will have an option to connect to a smartphone app to provide further convenience when perform- ing its function. The first problem with these uses of WiFi is that when the devices operate on the same channel that the airport network is using for an access point in the area, it degrades the signal-to-noise ratio and slows the network down. When such a device is close enough to either an access point or to a client device on the airportâs network to trigger the clear channel assessment (CCA) threshold, these other devices will back off, to allow it to use the channel. Of course under the CCA protocol, it should also be backing off for the airport network and its cli- ent devices. However, while the CCA protocol prevents direct interference, it impacts both the airport network and the independent users by reducing the time available to transmit. A second problem is created when these independent WiFi users are on an adjacent channel. WiFi devices vary greatly in how frequency-selective they are and some are sensitive to trans- missions that are many channels away from the one they are operating on. In this situation, the adjacent channel transmission will usually not be recognized as a WiFi signal and so will not
WiFi Service at Airports and the Problem of Interference 21 trigger the CCA protocol. With adjacent channel operation, the result is similar to what happens when non-WiFi transmitters are in the area. The transmissions create noise, slow the transmis- sion rates, and cause packet loss requiring retransmissions. When the CCA protocol is triggered, there is an impact to network performance, but the loss is controlled and planned to minimize the impact. When energy enters into a WiFi device from an adjacent channel, the interference is much worse because there is no mechanism to coordinate use. Network Design Solutions to Reduce Interference Radio frequency environments at airports are constantly changing and reflect the wide swings in network use by travelers who bring their own devices, as well as the new ways that airport operations take advantage of this technology. The challenge is to design a network that is equal to the challenges it will face and then integrate the tools and practices that will optimize the net- work in a timeframe that is meaningful, given the ongoing nature of the networkâs environment. Active Network Testing Many network administrators assume that all they can do is just respond to complaints. An alternative to resolving problems ad hoc is to actively test the network and monitor its perfor- mance. Active testing can identify a problem and potentially correct it before users even notice that a problem is occurring, whereas in passive testing performance is observed and reported, but nothing is done to change the network. There are as many as 600 different parameters that can be measured. Each one provides different information and insight to the networkâs perfor- mance. The kind of testing or monitoring that will best serve an airport network depends on its size, complexity, and applications running on the network. With passive testing, monitoring can be external to the network or it can be integrated into the network. There are advantages and disadvantages to each approach. If monitoring is done outside the network, the number of tools available is much larger. A number of devices are avail- able to do packet captures and each has particular strengths. If RF is the primary concern, then spectrum analyzers can be used to see what is happening in the RF environment. Increasingly, low-cost spectrum analyzers are coming to the market and they provide an excellent value for the performance they deliver. However, for some measurements only more capable instruments will make the needed measurement. All the major network equipment vendors are integrating tools for monitoring the network into their equipment and software. Integrated monitoring has the advantage of being built- in, portraying interference the way the network experiences it. Of course, that can also be a disadvantage. Some kinds of interference happen because the network is not able to monitor the sources but is impacted by them. To get a handle on those problems takes an external tool. Integrated network monitoring is the right tool for many purposes and situations, but other situations require tools that give an independent view of interference with data transmission. Active testing of a network involves using the network in some way and recording its perfor- mance. A very popular active test is to check the upload and download speed from a particular connection. Active testing gives better insight into network speed, latency, and capacity. Passive monitors can only observe what is happening with the data others are transmitting or receiving through the network. With active testing, the data load and way it is sent is under the control of the test. Like monitoring, active testing can be integrated into the network or it can use external tools. An example is a specialized access point used to simulate a client device and test the speed and
22 A Guidebook for Mitigating Disruptive WiFi Interference at Airports capacity of the access points near it. These kinds of specialized access points can be integrated into the regular network and create an independent testing sub-network. The goal of both monitoring and testing is to spot problems early or sometimes before they actually occur, with the objective to resolve problems very quickly or prevent them altogether. The right tools and methods will depend on the characteristics of a specific network, the train- ing and skill of the network management staff, and the service levels the network must support. With the increasing reliability expectations of modern networks, testing and monitoring are as essential as warning lights and periodic checkups of cars, planes, and other systems we rely on. Clear Channel Assessment Clear channel assessment (CCA) is an attempt by the WiFi system to determine if the trans- mit channel is busy or available before it attempts to transmit any data. If busy, it will estimate the duration for how the long the medium will be used before attempting to transmit. If a WiFi device is too sensitive to signals in an adjacent channel, that energy is included in its CCA. How- ever, because the signal is not operating on its channel the CCA will often not identify the device as a WiFi device and will transmit at much higher levels, potentially causing interference to that device and also receiving interference from that channel. The problem is that CCA assumes WiFi devices are capable of isolating transmissions on their channel from transmissions on other channels. However, for cost savings many WiFi devices do not have the frequency selectivity they need and are very sensitive to transmission on other channels. The lack of frequency selectivity is not important when there are no transmissions on other channels. However, in a crowded environment like an airport, good frequency selectivity is critical for devices to be able to operate simultaneously on different channels and for CCA to operate as intended for those operating on the same channel. A key is to install the appropriate number of access points, but to place them so that they are as isolated from each otherâs signals as possible. There is a big difference between installing an access point with its antenna pointed up through concrete arbitrarily and doing a seemingly identical installation with good data and understand- ing of the entire network design. It is undesirable to arbitrarily place an antenna out of sight because its appearance has been deemed architecturally unacceptable, without an understanding of how the placement affects coverage. It is better to choose such a placement after testing has assessed the loss introduced by the concrete and efforts have been made to accommodate that loss in the design. A particularly interesting network diagnostic application of directionality is to have one access point with multiple directional antennas, or an antenna array that allows beam formation, and then actively test the surrounding access points and monitor network performance. Using directional- ity, one access point can look in multiple directions. It can be used as a spectrum analyzer to look for sources of interfering RF. It can act as a client, actively connect to a neighboring access point, and test the data rate of that network node. Active, integrated network testing will be discussed in greater depth later in this Guidebook, but antenna directionality is an enabling technique. Placement of Access Points: Path Management The installation of WiFi access points must take into account the airportâs interior architecture and design. The path between the access points and users is critical to connection quality and net- work performance. The RF characteristics of walls, ceiling tiles, and other objects between an access point and its users have a profound impact on how well the signal gets to its intended recipient.
WiFi Service at Airports and the Problem of Interference 23 Most access points come with installation instructions and grounding planes, and are designed to operate from ceilings or elevated mounts. However, a common problem is for architectural or aesthetic considerations to conflict with optimal placement of access points. The person with architectural control may decree that the access point be placed out of sight, perhaps behind the ceiling tiles. If the impact of that decision on the RF signal quality is explored, there may be workable solutions. In a typical installation, an access point needs power and an Ethernet connection to the wired network. These are not always available in the best location for the access point. Availability of power and an Ethernet connection are legitimate concerns, as are architectural aesthetics. How- ever, network performance is also a legitimate concern. Measurements should be made during and after installation to support the placement decisions with good data on the impact to system performance. Complying with equipment installation instructions and working with the airport manager will help mitigate these potential problems. Distributing WiFi Traffic Crowded communication channels are harder to manage. Hence, one of the clear opportuni- ties for reducing WiFi interference is to distribute the WiFi traffic more evenly and then to opti- mize that distribution for the specific electromagnetic environment for problematic locations. To successfully distribute WiFi traffic requires action on both the network and the user sides. There are also several challenges to address, which are discussed in the remainder of this section. In the measurements made at airports, an unbalanced distribution of WiFi traffic was observed. Approximately 80% of WiFi traffic is crowded into the 2.4 GHz band, even though there are many more channels and much less interference in the 5 GHz bands. When looked at on a per channel basis, in many cases a majority of the traffic is concentrated in a single chan- nel. This was observed in both the 2.4 GHz and 5 GHz bands, where the traffic in the bands was concentrated into one or sometimes into two or three of the available channels. When the band utilization was compared to the number of transmission errors, there was a clear correlation. As a channel gets more congested, the number of transmission errors goes up. Optimizing for the Electromagnetic Environment At some locations, there are particular challenges. For example, a leaky microwave might be emitting strongly at 2.45 GHz. This would cause interference to the channels in the center of the 2.4 GHz band but not those away from the center. If the microwave cannot be removed, perhaps because it is owned and necessary for a tenantâs operation, then using channels that operate away from its emissions might be a solution. There are a variety of location-specific issues that can arise. Where the interfering source is fixed or frequently present at a location, one option is to plan the network around that reality. While it is preferable to remove sources of interference, that is not always possible. Therefore, an alternative is to plan the network so that the potential for interference is avoided. Optimizing for the electromagnetic environment is challenging but much more achievable than planning for a dynamic environment. However, with a bring-your-own-device environ- ment the electromagnetic environments are very dynamic. To address this, some network equip- ment manufacturers are building in the capability for the network to monitor and automatically configure itself to avoid interference. This concept is still very early in the implementation stage and will take some time to mature.
24 A Guidebook for Mitigating Disruptive WiFi Interference at Airports Dynamic Frequency Selection and Transmit Power Control Requirements Dynamic frequency selection (DFS) is the process of detecting signals (i.e., radar) that must be protected against WiFi interference, and upon detection switching the WiFi operating frequency to one that is not interfering with protected systems. Transmit power control is used to adapt the transmission power based on regulatory requirements and range information. Dynamic frequency selection and transmit power control are used in WiFi systems to change frequencies in order to avoid interfering with other users, as well as to keep the power as low as possible con- sistent with their ability to communicate and support their intended function. Transmit power control ensures the WiFi noise in the environment is no higher than needed for it to do its job. Although DFS channels have been made available by the FCC, there are challenges to be addressed. A random sampling of equipment found that it is relatively common for manufactur- ers to bring their equipment to the market without qualifying it for operation on the DFS channels. It was also found that most of the equipment surveyed came with the DFS channels turned off as the default setting. The user had to go into the configuration and manually turn on the channels. This was true of both access point and user equipment. To use the DFS channels, both the network operator for the airport and the traveler connecting to the network must turn on the DFS channels. The FCC requires a number of channels to implement DFS and transmit power control mea- sures to avoid the potential for interference with other radar systems that share their frequencies. The belief that the DFS channels should be avoided to protect radar is a misunderstanding of the FCCâs intentions. The FCC wants to see the spectrum utilized effectively. The DFS and transmit power control mechanisms are designed to give adequate protection to radar systems. Network designers do not need to avoid these channels. There have been problems with some manufac- turers or network designers disabling these features, with resulting interference to radars in areas where WiFi and radar were in close proximity. However, a function that is turned off cannot be judged to be ineffective. The ability to successfully use the DFS channels at airports is demonstrated at Dallas/Fort Worth International Airport (DFW), Logan International Airport (BOS), and Baltimoreâ Washington International Airport (BWI), which have Terminal Doppler Weather Radar sys- tems using DFS channels in their terminal WiFi networks. Figure 19 shows the current U.S. band plan for the unlicensed national information infra- structure bands and the band extensions being considered by the FCC. These DFS channels represent important opportunities to distribute WiFi traffic more evenly and reduce congestion and interference by simply moving to channels where there is no interference. In addition to 6 FCC 14-30, paragraph 4. Figure 19. U.S. band plan for the unlicensed national information infrastructure bands with dynamic frequency selection requirements shown.6
WiFi Service at Airports and the Problem of Interference 25 what is shown here, there are other bands used by WiFi: the new TV White Space band, and the 3.6 GHz, 4.9 GHz, and 60 GHz bands. Spectrum Reuse and Load Distribution Spectrum reuse and load distribution are important aspects of network planning. Spectrum reuse looks at the selection of RF channels so that access points do not interfere with each other. Load distribution seeks to even out the data load being processed to avoid bottlenecks or imbal- ances in the network. There are relatively few WiFi channels available. The network should be planned so that each access point has full use of its channel with a minimal potential for interference from neighbor- ing access points on the same or an overlapping channel. Spreading out the RF channels is a simple concept but can become complex to implement well. Similarly, load distribution is a simple concept. Ideally the various communication streams would be spread out among the various communications paths so that the processing power of the network is being utilized optimally. It can be challenging to avoid imbalances when some access points or specific channels in an access point are overloaded and experiencing problems, while others are underutilized. In an optimized network, the communication sessions will be spread among the available RF channels and data paths in ways that make best use of network resources. However, this ideal is seldom observed in operating networks. Measurements at airports show a high percentage of the traffic being crowded into the 2.4 GHz band and, within that band, into one of the available channels. This leads to congestion, WiFi-to-WiFi interference, high error rates, and a variety of network problems. Access Point Channel Distribution Poor channel assignments can create a number of problems in networks. Increasingly WiFi network control software will automatically change the channel assignment of access points in an effort to separate them in frequency. However, under some circumstances the software can do just the opposite of what it is designed to do and put neighboring access points on the same channel, which results in interference between them and their client devices. The same kind of problem can be created when the channels are assigned manually without taking into account the importance of separating neighboring access points in frequency. Figure 20 shows a poor distribution of WiFi channels, with channel 1 being used in a number of adjacent access points. Such a design is likely to result in interference problems. Figure 20. An example of poor spectral reuse. Ch 11 Ch 1 Ch 1Ch 6 Ch 1Ch 1 Ch 1 Ch 1
26 A Guidebook for Mitigating Disruptive WiFi Interference at Airports Table 4 shows the frequency assignments at a gate in one of the largest U.S. airports. Notice that the network designer tried to distribute the access points among the non-overlapping channels in the 2.4 GHz band. There are two access points using channel 1, which might be problematic. However, in the layout shown in Table 4 the real problem is from the non-network access points, which are using the same or an overlapping adjacent channel. There is a strong potential for these access points and the clients that connect to them to have significant adjacent channel interference, with the resulting degradation in network performance. In a good network design, access points that use the same channel will be distributed to have channels with the same fre- quency spaced as far apart as possible. Automated Network Management In the past, radio systems were relatively rare by todayâs standards, seldom used, and often in a fixed location. Managing radio systems of the past could be done with paper and pencil and plans developed in meetings among the users and managers of the systems. However, todayâs wireless systems are everywhere and in constant operation. Networks and devices are in regular communication, with no user intervention involved. Many services like automatic e-mail alerts are activated when devices have data to deliver; no person is involved in the communication session. Given the very dynamic environment in which airport WiFi networks operate, manual methods cannot keep up. WiFi networks at airports are faced with a constantly changing envi- ronment in which the load placed on them and the competition for spectrum from other WiFi devices and non-WiFi devices are in a constant state of flux. Only by embedding into the net- work the ability to sense its environment and adjust in near real-time can there be a possibility of keeping interference at a minimum and network performance at target levels. That is what automated network management accomplishes. There are a variety of software and hardware options available for automating network man- agement tasks that previously had to be done manually. Newer versions of the WiFi standards support multiple modulation and coding states that allow a wide range of data transmission rates. When signaling conditions are good and competing traffic is light, a user can experience superb transmission speeds. However, as the signal degrades, perhaps because the user is farther WiFi Channel Airport Network Access Points Non Network Access Points 1 2 4 2 1 3 4 5 1 6 6 7 8 9 1 10 11 4 Table 4. An example of an airport gate with significant potential for adjacent channel interference.
WiFi Service at Airports and the Problem of Interference 27 away or because of interference from other transmitters, the system backs off, dropping to a slower speed but more reliable modulation methods. Some systems will track the availability of other access points and switch access points when the signal from one becomes better than the current connection. However, network automation is in its infancy, and due diligence and per- formance monitoring are essential to successfully using it. For the airport manager concerned with network performance, the relationship with the network manager may need to be more hands-on than in the past. Key tools used by network automation are changing channels and adjusting power levels. The objective is to get access points assigned to channels and operating at power levels that provide maximum performance with minimum interference. However, the automation algo- rithms have shown a tendency to misallocate both channels and power levels. The result can be an unstable system. There is no option to return to the past, slow, manual methods of network management. New automatic network management tools are under active development, but they are far from achieving all that is needed from them. Manufacturers must sell todayâs products to raise the revenues to improve them. Network managers must practice due diligence and equip themselves with the tools and expertise to manage their networks just as they would any other complex system. A repeated problem area is when equipment manufacturers design equipment for use with a particular network architecture, and then the network designer uses different design architec- ture. Equipment and functions that work well in one context may become more of a problem than a solution when operating in a network design that its manufacturer did not anticipate. For example, a method used in automated network management is to automatically adjust power levels downward to keep access points from interfering with each other. Often the decision is made by measuring the signal level of neighboring access points and adjusting the power so that the signal level is below a predefined limit. However, unless client connectivity and area cover- age are appropriately considered, the result can be performance issues and a poor client experi- ence. The network design may need access points to have a certain signal strength to adequately reach into remote parts of the facility. Turning the power down may minimize interference between two access points at the expense of providing coverage to the entire facility. Sometimes compromises are needed and the automated algorithms are not always equipped to understand conflicting needs. Network automation has a growing role in managing the ever-increasing complexity and demand placed on networks. However, those responsible for network performance need to be vigilant in its use, particularly when it is new.7 Channel Assignments One technique that has proven useful is to look at the channel plan on the networkâs manage- ment console. The channel numbers need to be superimposed on an area map. The purpose is to analyze the distribution of channels. Are all of the allowed channels being used? Are the channels evenly allocated within the area? Are neighboring access points using different channels? The check of channel assignments should initially be repeated several times a day to look for unnecessary churn. Proper channel allocation is a key element of achieving good network per- formance with minimal interference. 7 For further discussion of the topic, see: Veli-Pekka Ketonen, âWLAN Access Point Automation Issues: What You Can Do,â 2014. Available at: http://7signal.com/blog/wi-fi-access-point-automation-issues-what-you-can-do/
28 A Guidebook for Mitigating Disruptive WiFi Interference at Airports Next, the power levels and field strength need to be checked over the coverage area. Power levels at the low end of the power range are likely to result in poor coverage and decreased performance. Cases have been reported where access points are configured by automated sys- tems to operate close to 0 dBm. Such low power levels are unlikely to be consistent with good performance. HotSpot 2.0 Another capability to automate the management of WiFi networks is the use of HotSpot 2.0. HotSpot 2.0 originated in the standard IEEE 802.11u and is designed to allow automatic con- necting for all of its users. It is viewed as the secure and easy option for roaming with cellular data.8 The idea is for mobile users to be able to automatically join WiFi subscribers whenever the user enters an area of coverage. The key elements are network discovery and selection, stream- lined network access, security, immediate account provisioning, and provisioning of operator policy for network selection. Another improvement is integration with WiFi Passpoint, a pro- gram certifying that access points and devices comply with technical specifications.9 HotSpot 2.0 would alleviate a great deal of the congestion currently occurring, as well as provide faster speeds for mobile users. Distributed Antenna Systems A distributed antenna system (DAS) is a network of spatially separated antenna nodes con- nected to a common source via a transport medium that provides wireless service within a geographic area or structure. Distributed antenna system elevations are generally at or below the clutter level, and node installations are compact. Use of DAS allows the network manager to split the level of power transmitted over an area by spreading the transmission over multiple antennas versus one antenna. The net effect is to reduce the possibility of interference to local users as well as increase reliability. 8 Ruckus Wireless, âHotSpot 2.0: Making the Wi-Fi Roaming Experience as Secure and Easy to Use as With Cellular Data.â Available at: http://www.ruckuswireless.com/technology/hotspot2 9 vonNagy, Andrew, âWiFi Alliance Rebrands Hotspot 2.0 as WiFi Certified Passpointâ 2012. Available at: http://www.revolution wifi.net/2012/05/wi-fi-alliance-rebrands-hotspot-20-as.html
