Preparing your wlan_wp_0409_chv3

384 views

Published on

Published in: Business, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
384
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
5
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Preparing your wlan_wp_0409_chv3

  1. 1. WHITE PAPERPreparing your WLANinfrastructure for voice
  2. 2. IntroductionAdding mobility to enterprise voice communications can be the key to unlockingbusiness productivity and responsiveness. Whether addressing productionproblems, planning promotions or responding to customer requests, enterprisesrequire that associates be able to move about freely and still be accessible at amoment’s notice. Those communications links must extend from the office tothe shop floor and out to the loading dock.The Webtorials 2008 State of the WLAN Report indicates that 36% of enterprisewireless LANs support voice today, and an additional 33% plan to add itwithin the next 12 months. Organizations of all types are using their wirelessLANs to support mobile workers with both connection-based and push-to-talkvoice services in a wide range of applications. From healthcare to retail andwarehouse management, users are discovering the benefits of cost-effectivecontinuous access. However, that flexibility and responsiveness will depend onhaving a reliable mobile network that allows employees to remain productivewhile they move through their daily activities.A sound and reliable wireless infrastructure is needed to ensure the qualityand availability of a mobile voice service. At the outset, it is imperative that thenetworking group recognize the requirements of a WLAN voice network, assessthe capabilities of their existing infrastructure and take the necessary steps tobring the network capacity and coverage up to the levels needed to supportenterprise-quality voice services. Mobility will not lead to improved productivityif the network is unreliable or unable to provide good voice quality.
  3. 3. Defining requirements Getting your wired networkfor WLAN voice ready for WLAN voiceThe first step in a voice over WLAN (VoWLAN) The backbone for your WLAN voice service willproject will be to develop a definition of service you be the wired LAN. Whether your wired voicelook to provide, and that starts with an adequate system is based on an IP PBX, a traditionaldefinition of requirements. The first step is to TDM model or a hybrid configuration, you willspecify the number of users to be served, the type have to interconnect calls between the wiredof voice service you will provide (i.e., traditional and wireless systems. Further, if you will bevoice calling or push-to-talk), the areas where the supporting push-to-talk (PTT) devices, theyservice will be available and the expected traffic should be able to interconnect to any other PTTvolumes. Having a handle on traffic volumes and systems or services you already have in place.usage patterns will not only help in planning yourinfrastructure requirements, it will also be key in The requirements for the wired network intercon-determining other parameters such as battery nection will vary based on the nature of therequirements. wired telephone system and the signaling that is used on the wireless LAN voice devices. AsIn contrast to data users who typically operate shown in Figure 1, if the wireless LAN handsetsfrom a stationary location, voice users are highly use a signaling protocol that is compatible withmobile, so it will be difficult to predict accurately the telephony server on the wired IP PBX, youwhere they will be when they need to make or should be able to pass calls directly throughreceive calls. That means you will also require a to the wired LAN. On the other hand, if yousound network management system that allows are using a TDM PBX or a WLAN handset thatyou to identify capacity and coverage problems, employs a proprietary signaling system, you willand plan for expansion. Further, mobility is highly require a gateway between the wireless LANappealing, and as other employees see that the devices and the wired PBX. In the longer term itservice is available, you can anticipate more is expected that all voice services will migrate torequests, more mobile handsets and hence more IP technology and the Session Initiation ProtocolWLAN voice traffic. (SIP) will be adopted as the signaling standard, which should greatly ease the integration ofIn defining your requirements it will also be wired and wireless users.important to categorize the various types of usersto be supported (e.g., general office, tech support, If you are using an IP PBX or a hybridsecurity, production, etc.), the criticality of their configuration that supports LAN-connected IP/communications (e.g., general business calls Ethernet handsets, there are several features andversus security or emergency services) and the configurations that are universally recommendedtypes of handsets or other mobile voice-enabled for the wired LAN infrastructure. Those woulddevices they will be using. Classifying users can include:help to quantify the volume and location of callingand will also be useful in predicting the amount • A fully-switched LAN configurationof traffic additional users of that type will likely (i.e., no hubs)generate. Finally, you should identify the types • Use of 802.1p QoS for prioritizing voiceand models of the WLAN voice devices you will frames over the LANsupport, and the systems for maintaining them. • Power over Ethernet (PoE) to power wiredNew WLAN voice devices are introduced regularly, IP handsets and WLAN access pointsso you should define the procedure by which • All voice devices should be configured on anew devices are tested before they are added to separate virtual LAN (VLAN) for improvedthat list. security capabilities3 WHITE PAPER: Preparing your WLAN infrastructure for voice
  4. 4. Figure 1: If the wireless LAN handsets use a signaling protocol that is compatible with the telephony server on the wired IP PBX, youshould be able to pass calls directly through to the wired LAN. On the other hand, if you are using a TDM PBX or a WLAN handset thatemploys a proprietary signaling system, you will require a gateway between the wireless LAN devices and the wired PBX.These features must be coordinated with the 54 Mbps 802.1a or g radio links require a 100 MbpsWLAN voice implementation. The WLAN is a connection to the wired LAN. If you plan to upgradeshared media configuration, but the use of a fully- to 802.11n, the access points will require 1 Gbpsswitched wired infrastructure will minimize wired connections. WLAN voice handsets supportingthe delay for delivery of voice frames over the 802.11n are not likely to appear for the next fewwired network. The 802.1p QoS standard is years, but n-capable access points might still beimportant to minimize delay in forwarding voice used to provide higher capacity data services, soframes, and that implementation will have to be choosing a vendor with a strong 802.11n portfoliocoordinated with the WLAN’s 802.11e QoS; those will be key.issues will be described later. To associate WLANvoice devices with the voice VLAN in the wired Finally, the network management capabilities ofnetwork, you will have to define separate wireless the wired LAN network should be investigated withVLANs with different network names (i.e., SSID’s) regard to their ability to provide information that willfor the voice and data traffic. Those wireless voice be useful in supporting voice applications.and data users may be sharing the same WLANchannel, but each group can still be associatedwith the appropriate wired VLAN. The WLAN infrastructure The basic requirement for a WLAN infrastructure toThe WLAN access points will be connected over support voice is dense, pervasive coverage. Densitythe wired infrastructure. For that, it is important refers to the signal strength and pervasivenessthat the LAN switches be capable of supporting refers to the coverage. Signal strength impacts thethe required number of IEEE 802.3af Power over transmission rate users receive on the networkEthernet (PoE) ports. Both wireless LAN access and hence the number of simultaneous calls thatpoints and wired IP voice handsets use PoE, so if can be supported on an access point. The generallyyou are using or planning to migrate to an IP PBX, accepted design parameter is a received signalyou will want to ensure there is sufficient PoE strength floor of -67 dBm, though better designedcapacity on your LAN switches or be prepared to handsets can often work down to -70 dBm. The goalinvest in mid-span PoE devices. Further, you must is to provide signal strength that will result in theensure that the power supplied is sufficient for the most efficient network utilization, the shortest transitaccess points you intend to use. delays and the maximum number of calls supported.One major development in WLANs is the With regard to density, there are two importantintroduction of the new higher capacity 802.11n factors that characterize WLANs: shared media andradio link. WLAN access points supporting the adaptive modulation. Shared media means that4 WHITE PAPER: Preparing your WLAN infrastructure for voice
  5. 5. all devices associated with an access point take potential 23 non-interfering channels versus threeturns using one half duplex channel. As with any in the 2.4 GHz band. Further, the 2.4 GHz channelscontention-based network, the greater the volume might already be congested with data traffic,of traffic vying for access to the channel, the greater particularly if 802.11b and g devices are sharingthe delay that users will experience. Good signal the channel. The 5 GHz band provides an expansivecoverage results in better network efficiency, and frequency window for voice deployments withthat in turn leads to lower transit delays, a key factor fewer interference issues and no impact on 2.4 GHzin providing high quality voice. For enterprise-grade data users.voice service, the requirement is to provide one-way,end-to-end delay below 150 msec. The key to providing high-quality voice service is a WLAN network design that delivers good signalBetter signal coverage also leads to higher strength throughout the desired coverage area;transmission rates. WLAN devices use adaptive everything works better with a strong signal.modulation, which means the WLAN device Devices will transmit at the highest data rates, thereduces its transmission rates as the signal strength channel will be used more efficiently and theredecreases and the signal-to-noise ratio degrades; will be fewer retransmissions all of which leads tothe range of data rates supported on WLANs is higher call capacity. Good signal quality is a resultsummarized in Table 1. Signal strength is primarily of sound network design and a configuration witha factor of the distance to the access point and any sufficient access points to support the expectedmaterial obstructions in the path. In a shared media volume of voice traffic. Maintaining that levelnetwork, adaptive modulation means that faster of performance over time requires a networkand slower transmitters will be sharing the same management system that monitors traffic volumeschannel. It stands to reason that the channel will be and identifies problem areas before they affect userused most efficiently if all stations transmit at their performance.highest data rates. Further, devices with poor signalquality will not only transmit at lower rates, they will With regard to coverage, most organizationshave to retransmit more frequently, increasing delay have deployed WLANs with “spot coverage”and degrading efficiency. in conference rooms, public areas and other defined areas (e.g., loading dock, warehouse,In selecting WLAN voice equipment it is important etc.) where they need to support mobile devices.to locate devices that can operate in both the The Webtorials 2008 State of the WLAN Report2.4 GHz (i.e. 802.11b/g) and 5 GHz (i.e.802.11a) notes that only 55% of users report having WLANbands. Support for the 5 GHz 802.11a interface coverage throughout the office areas. Voice usersprovides far greater flexibility in the network may wander anywhere within the facility, and theconfiguration. First, the 5 GHz band provides a WLAN service will have to be available so they can make and receive calls. IEEE 802.11 Radio Link Interfaces Non- Max. Fallback Channel Transmission Radio Standard Interfering Bit Rate Rates Bandwidth Band Technique Channels 5.5 M, 2 M, 02.11b 11 Mbps 22 MHz 2.4 GHz 3 DSSS and 1 Mbps Same as 802.11a plus 02.11g 54 Mbps 20 MHz 2.4 GHz 3 OFDM 11 M, 5.5 M, 2 M, 1 Mbps 48 M, 36 M, 24 M, 18 M, 02.11a 54 Mbps 20 MHz 5 GHz 23 OFDM 12 M, 9 M, and 6 MbpsTable 15 WHITE PAPER: Preparing your WLAN infrastructure for voice
  6. 6. It is generally accepted that any large-scale, repeated on a smaller scale each time a new accessenterprise-grade wireless LAN should be built point was added to the network.using a centrally controlled WLAN switch. AsWLANs grew in size and importance, it became One of the important developments in wirelessclear that networks built on autonomous LANs has been the introduction of computerizedstandalone access points were too difficult to network design tools. To use the design tool, thedesign and manage. In a centrally-controlled user first imports a CAD drawing of the facility. Theysolution, a network of thin access points is then define the scale, building materials (e.g., sheetcoordinated by a central controller that can assign rock versus cinder block walls) and furnishings aschannels and adjust transmit levels automatically to they will affect the signal propagation. Finally theyensure good coverage throughout the area. identify the number of users, capacity requirements and whether the network will be using 2.4 GHz orThe major decision regarding the infrastructure will 5 GHz channels. The better systems also take intobe whether voice and data devices are supported account the design of the handset, in particularon the same or on different wireless LANs. While the antenna. In that way it is possible to provide anthe idea of building a separate WLAN for voice accurate assessment for both inbound and outboundwas originally viewed as wasteful extravagance, transmissions. Based on those inputs, the toolcentrally-controlled WLAN switches are making generates a design for the installation that identifiesthis strategy more cost effective. A single WLAN the number and placement of access points, thecontroller can typically support both networks, channels to be used in each area and the transmitand many commercial access points can be power setting based on formulas that reflect signalconfigured with two radios. As a result, much loss based on frequency, distance and materialof the infrastructure can be shared. This type of obstructions.deployment is called a dual overlay network, and itwould typically use a 2.4 GHz 802.11b/g network The result is that you can have a highly accuratefor data devices and a 5 GHz 802.11a network design that can cut weeks off the time it takesfor voice. It is important to note that signal loss to tune the network. Rather than a set of circularis greater at 5 GHz than at 2.4 GHz, so a 5 GHz coverage areas centering on each access point,network will typically require more access points to you will typically find that the building materialseffectively cover the same area. and other features in the environment shape the coverage area. With a sound preliminary design, the RF management capabilities of a centrally controlledNetwork design tools WLAN switching system will allow you to implementA sound network design is the starting point for a network that is capable of supporting the stringentany voice-capable WLAN infrastructure. The first requirements of WLAN voice traffic.generation of wireless LANs were built using aninexact and time-consuming process of trial and WLAN network features for voiceerror. That process involved conducting a sitesurvey, identifying potential locations for access While a sound radio infrastructure will be essentialpoints, assigning channels to each and then for any WLAN voice deployment, there are severaladjusting the transmit power to achieve adequate specific features that will also be important for voicecoverage with minimal interference between support. In particular, these features will deal withaccess points assigned to the same channel. With quality of service (QoS), handoffs and battery life.only three non-interfering channels in the 2.4 GHzband, minimizing interference could be highly IEEE 802.11e/Wi-Fi Multi-Media (WMM) Qualityproblematic. of Service (QoS) To recognize the requirement for WLAN QoS, it isOnce the preliminary installation was complete, important to understand a little about the WLANthe network designers could then spend a access protocol used on wireless LANs. WLANsconsiderable amount of time tuning the network. use a protocol called Carrier Sense Multiple AccessThose adjustments would involve relocating with Collision Avoidance (CSMA/CA). Wirelessaccess points to improve coverage, reassigning LAN stations transmit and receive on the samechannels and adjusting transmit power to minimize channel, so, when a device is transmitting, it cannotinterference. That process would have to be hear other transmitters; hence there is no way to6 WHITE PAPER: Preparing your WLAN infrastructure for voice
  7. 7. “detect” collisions as is done in a traditional wired The original 802.11 CSMA/CA protocol defined twoEthernet. To complete each transmission, the pre-transmission waiting intervals:receiving station tests the frame for errors andreturns an acknowledgement. • DCF Inter-Frame Spacing (DIFS): The interval a station waits before sending a frameAs collisions cannot be detected, the WLANprotocol takes steps to help avoid collisions. • Short Inter-Frame Spacing (SIFS): The intervalWhen a WLAN station senses that the channel a station waits before sending anis idle, it waits a defined interval called an Inter- acknowledgement or ACK.Frame Spacing before it attempts to transmit. Ifa collision or other failure occurs (i.e., the sender As the SIFS interval is shorter, if one station isdoes not receive an acknowledgement), the waiting to send a frame and another is waiting tostations back off by a random interval before trying send an ACK, the ACK will always be sent first.again; that back-off range is called a contentionwindow (CW). The stations also back-off a random Recognizing the time-sensitive nature of voiceamount if they sense the channel is busy when transmissions, in 2005 the IEEE introduced athey try to access it. quality of service mechanism designated 802.11e; the Wi-Fi Alliance identifies products that are compatible with that standard as Wi-Fi Multi-Media (WMM) Certified.Figure 2: To help avoid collisions, the WLAN access protocol uses a system of pre-transmission waiting intervals.Those waiting intervals are of varying duration, so they can serve as a mechanism for prioritizing transmissions (e.g.,higher priority transmissions are assigned shorter intervals). If collisions occur, the intervals can be random to lessenthe likelihood of subsequent collisions. The above diagram shows the process of transmitting a frame and the resultingacknowledgement. The acknowledgement is sent with the shortest waiting interval (i.e., SIFS), which means that theacknowledgement will be sent before any other traffic can be generated.7 WHITE PAPER: Preparing your WLAN infrastructure for voice
  8. 8. The 802.11e EDCA/WMM option defines an The four priority levels or Access Categories (ACs)“enhanced” access mechanism with different are designated:pre-transmission waiting intervals called ArbitratedInter-Frame Spacing (AIFS). To give time-sensitive AC 1: Voicevoice and video transmissions higher priorityaccess to the shared radio channel, they are AC 2: Videoassigned shorter pre-transmission waiting intervals.So if a voice user and a data user are both waiting AC 3: Data: Uses the same pre-transmission intervalto transmit a frame, the voice user will always go and back-off range as legacy (i.e. pre-802.11e)first. The standard also defines shorter back-off WLAN devicesranges (i.e. CWMIN and CWMAX) for the higherpriority retransmissions. AC 4: Background Data The AIFS and Contention Window ranges for each access category are summarized in Table 2. IEEE 802.11e Default Parameters Parameter DSSS PHY (802.11b) OFDM PHY (802.11a/g) Access Category OFDM Access Category (802.11a/g) 1 2 3 4 1 2 3 4 IFS (SIFS + x 2 2 3 7 2 2 3 7 Time Slots) IFS Time 50 50 70 150 28 28 37 73 (µsecs) WMIN 7 15 31 31 3 7 15 15 WMAX 15 31 1023 1023 7 15 1023 1023Table 2Note: Time slot duration is 20 µsec for 802.11b and 9 µsec for 802.11a/g8 WHITE PAPER: Preparing your WLAN infrastructure for voice
  9. 9. Given the shared media design of a WLAN, Battery life802.11e/WMM will be a critical element in One last though critical element in providing anensuring enterprise-grade voice services over enterprise-grade voice over WLAN solution hasthe wireless LAN. The priority setting in the been battery life of the mobile device. Where cellWLAN must be coordinated in the access point phones routinely deliver several hours of talk timeconfiguration so that voice frames are marked and dozens of hours of standby operation on awith the corresponding 802.1p priority before single charge, early Wi-Fi voice devices providedthey are forwarded over the wired LAN. That QoS a fraction of that. The problem is that powermechanism must also be coordinated with the IP conservation was not one of the primary goals inDifferentiated Service (DiffServ) priority, as users the original Wi-Fi standards that were geared formay roam between different IP subnets. devices like laptops that could include large, bulky batteries.HandoffsAlong with QoS, a WLAN voice network must The original Wi-Fi standards did include a Poweralso be capable of handing off connections from Save feature, but it was not particularly effectiveaccess point to access point quickly and securely and introduced considerable latency for voiceas a user moves through the coverage area. In a transmissions. A far more effective power savingtypical WLAN voice deployment, the radius of an feature is included with the Wi-Fi Multimediaaccess point’s coverage area will be roughly fifty (WMM) QoS standard. Designated WMM-feet. At typical walking speeds, a user will cross Automatic Power Save Delivery (APSD), this featurethe coverage area of a cell in 20 to 30 seconds, allows for far more efficient power conservationso a call may experience several hand offs if the along with reduced latency for voice. For example,user is walking. The initial laptop oriented Wi-Fi the Session Initiation Protocol (SIP), the emergingstandards provided a handoff that might take 5 standard for VoIP signaling, involves considerableto 10 seconds. While that might be suitable for “chatter” between the end devices and thedata applications, it certainly does not meet the telephony server. By monitoring that traffic theperformance requirements of a voice application. access point can determine if it really needs to be forwarded over the radio link. As those advancedThe IEEE 802.11r committee has developed features are not yet defined in the standards, thea standard for fast, secure hand-offs, with a handsets and infrastructure elements must comeperformance objective of 50 msec for the handoff from the same vendor in order to implement them.time. Even without this standard, existing WLANswitching systems can provide handoffs in time Wi-Fi voice securityranges that are almost as good. Current systemssupport handoff latency between 10 and 150 Security is always an issue with telephone calls,msec; typically the longest intervals are for and it was certainly a concern with early WLANhandoffs that involve moving stations between voice networks given the security deficiencies ofIP subnets. However, even a 150 msec handoff the Wireless Equivalent Privacy (WEP) securityinterval will result in a barely perceptible click mechanisms. Fortunately those issues have nowin the conversation path. As time goes on, it is been addressed, and it is possible to provideanticipated that WLAN infrastructure vendors will security as sound as that typically found in publicall migrate to the 802.11r standard; however the cellular networks.ability to do fast, secure handoffs should not be adeterrent to deploying WLAN voice systems today. There are two major areas to consider when addressing security: device authenticationThe 802.11r standard will provide an improved and privacy. If the authentication system ishandoff function. Using the IEEE 802.11k standard compromised, unauthorized devices could makefor Radio Resource Management, the Wi-Fi clients and receive calls over the network. That couldcan collect information regarding nearby access expose the network to toll fraud or theft ofpoints, a capability called neighbor reporting. Not service (i.e., paying for a hacker’s phone calls), callonly will that information be important to guide redirection and potentially registration hijackinghandoff decisions, it will also allow stations to do where an attacker is able to impersonate aopportunistic key caching, where they can store legitimate party. If the privacy mechanisms wereencryption keys for those adjacent access points. compromised, unauthorized parties would have theThe combination of 802.11r and 802.11k will allow ability to eavesdrop on WLAN phone calls.a station to roam to another access point morequickly as it will not need to secure an encryptionkey as part of the handoff process.9 WHITE PAPER: Preparing your WLAN infrastructure for voice
  10. 10. Today Wi-Fi voice devices typically use and attackers may also use them as part of aauthentication mechanisms based on the IEEE strategy to learn valid user names and passwords802.1x Extensible Authentication Protocol (EAP). they could use to access the network. Locating andThe most secure WLAN voice solutions utilize disabling those access points in a timely fashion willclient certificates, making such an attack virtually be critical in maintaining the security of the network.impossible. With a client certificate that is bound The WLAN security system should include theto the device’s MAC address, the Transport Layer ability to continuously monitor the RF environmentSecurity (TLS) protocol can forward the device’s to discover those security threats. When a rogueunique credentials in a secure, tunneled connection access point is located, the system should alert theall the way from the mobile device to the network managers, disable the unauthorized deviceauthentication server. and provide location information so that it can be found and removed.Eavesdropping on WLAN voice conversations isa potential concern, but only if the encryption isbased on the early Wired Equivalent Privacy (WEP). WLAN voice network capacityMost Wi-Fi voice devices today support 802.11i, Probably the most difficult issue to quantify inwhat the Wi-Fi Alliance terms Wi-Fi Protected a WLAN voice deployment is the number ofAccess-2 (WPA2) Certified. WPA2 uses encryption simultaneous calls an access point will be ablebased on the Advanced Encryption standard, the to support without degrading the voice quality ornew encryption standard for the US government noticeably increasing the transit delay. There are aunder Federal Information Processing Standard number of factors that contribute to the complexity197 (FIPS 197). In enterprise environments with of this problem, starting with the fact that the802.1x authentication, the authentication process amount of capacity required per call can be reducedproduces the encryption key and the solution has through use of voice compression. The ITU’sno known flaws. G.729A compression algorithm for example can reduce the voice payload from 64 Kbps to 8 Kbps.If WPA2 is not an available option, the Wi-Fi It is important to note that even though the voiceAlliance’s earlier WPA (Wi-Fi Protected Access) payload is reduced by a factor of eight, the overheadsolution can be used. WPA uses the same associated with WLAN voice will not accommodateencryption algorithm as WEP but a longer key , an equivalent number of additional voice calls.is used and the key is changed on every packet,effectively thwarting the type of brute force attacks Another factor that makes it difficult to specify thethat rendered WEP ineffective. When used with maximum number of calls the WLAN can support802.1x for authentication and key generation, what is the fact that different users may be operating atthe Wi-Fi Alliance calls WPA Enterprise, there are different data rates. The lower rate users will takeno known flaws. proportionally longer to send their voice frames, tying up the network for longer intervals and causingWhile authentication and privacy are the major other users to defer their transmissions. Assumingconcerns regarding user devices, wireless security a 50% maximum throughput on the network, themust also address the vulnerability of the network approximate maximum number of simultaneousitself. Users could connect unauthorized or “rogue” calls for different voice coding systems and averageaccess points, weakening the security perimeter, transmission rates is listed in Table 3. Approximate Maximum Calls Per WLAN (20 msec Voice Sampling, No Voice Activity Detection) 802.11b Network 802.11a or g Network Codec 11 Mbps 5.5 Mbps 2 Mbps 1 Mbps 54 Mbps 36 Mbps 18 Mbps 6 Mbps G.711 12 8 4 2 39 35 25 12 (64 Kbps) .729A 15 12 7 4 46 43 37 22 (8 Kbps) .723.1 22 18 10 6 69 64 55 33 (5.3 Kbps)Table 310 WHITE PAPER: Preparing your WLAN infrastructure for voice
  11. 11. A third element that could impact the voice call Given the dynamic nature of voice usage and thecapacity is voice activity detection (VAD), the relatively limited number of channels an accessoption of having voice packets sent only when the point will be able to support, users will be wellparty is actually speaking. Given the difficulty of advised to pilot test their planned configurationscoordinating that function with a wired PBX, most before progressing to a full scale roll out. Further, aWLAN voice systems do not use VAD today. The comprehensive network management system thatexception to that would be push-to-talk systems, will monitor voice usage, identify problem areaswhere voice packets are generated only when the and help plan for growth and expansion will betransmit key is depressed. critical to ensure that the network can continue to provide enterprise-grade service as the number ofPush-to-talk presents a different set of capacity users and the traffic volume grows.concerns. Used in a growing number of workenvironments, push-to-talk (PTT) over the wireless Network planning, trafficLAN can provide a more cost-effective alternativeto traditional walkie-talkies. As a PTT system monitoring and networkgenerates WLAN traffic only when the transmit managementkey is depressed, it should be inherently moreefficient than traditional voice services that Providing WLAN voice is not simply a matter ofgenerate frames continuously. However, in a picking some handsets, ensuring that they adherepoorly designed PTT solution, those PTT packets to a checklist of standards and passing them out.are broadcast through every access point, thereby The primary responsibility of the IT department is tocreating unnecessary WLAN traffic throughout the deploy a solution that will provide the basic service,entire network. An efficient PTT system should support the required features and have the toolsbe able to track users within the network and that are necessary to maintain and operate thetransmit their frames only on the WLAN where network. Those functions will be critical issues inthey are actually located. selecting the equipment needed to implement the solution.One last factor that comes into play if a sharedvoice/data network is deployed is the amount Earlier we introduced the computerized toolsof capacity to reserve for voice users. This is available to help design a voice-capable wirelessparticularly important when 802.11e/WMM QoS LAN. However, those tools can only help determineis implemented. As the QoS capability gives voice how to select and configure the equipment neededusers preferred access to the channel, if too many to provide the desired amount of wireless networkvoice calls are in progress, the data users could be capacity. The real design involves determining howsqueezed out entirely. So in a shared voice/data many users you will have, how much capacityWLAN, you must reduce the number of voice calls their calling volume will require and where theyan access point will accept to help ensure there will be when they need to make or receive a call.will be capacity available for data users. Common sense can provide some of that planning information. For example, if you have 200 peopleCall admission control (CAC) is the parameter that with WLAN handsets in an auditorium, you candefines the maximum number of simultaneous anticipate a torrent of voice calls as soon as thecalls supported on each access point; that meeting adjourns.parameter must be set in the access point or thecentral network controller. Some systems now Much of the necessary information for determininghave the ability to override the setting in the event WLAN voice capacity requirements can onlythat a user who is involved in a call roams into be gleaned from monitoring the actual networkthe area. In that case, a user who is involved in a utilization. That is where network managementcall might be allowed to roam in even though that systems become critical. Network managementwould technically push the number of calls over involves all of the systems required to help ensurethe defined parameter, while a new call request in delivery of a reliable and cost-effective service onthat area would be denied. an ongoing basis. The key element in that definition is “an ongoing basis. ”11 WHITE PAPER: Preparing your WLAN infrastructure for voice
  12. 12. The first step in providing good WLAN voice Traffic Monitoring. A mechanism will be needed toservice will be the ability to confirm that users determine if there is sufficient network capacity tocan get a usable signal to make and receive a call accommodate normal and peak usage in all areas.wherever they move within the facility. Once the Key to that will be the ability to identify the averagesignal is provided, it will be necessary to ensure and maximum number of users per access point,that there is enough network capacity in that area the periods of heaviest activity and the number ofto support the required volume of calls. That will call requests that are being denied. If voice andrequire some amount of coverage overlap among data are supported on the same network, it willadjacent access points. In short, some important also be necessary to gauge the impact of heavyassumptions will be made in the design of our voice traffic on data users.coverage plan, and if guessed incorrectly, someusers will be getting busy signals. Call Quality. The network might allow a user to make or receive a call but then does not have theUnpredictability is a given in mobile network ability to maintain the voice quality. VoIP qualitydesign, so network management systems that assessment tools are become a standard additionallow us to recognize and adjust to changing to IP PBX systems, however, the use of a wirelessconditions are a necessity. Further procedures LAN adds additional complexity to the problem.will be needed to deal with lost, stolen or broken The contention-based nature of wireless LANs willhandsets; terminated employees; handset typically increase transit delay and jitter, and cansoftware upgrades; equipment failures; areas potentially cause packet loss if the delay exceedswith poor signal coverage and all of the day-to-day the jitter buffer’s ability to compensate. Further,issues that go into providing a communication those parameters can vary widely during the call,service. particularly if the call is handed off access point to access point. You will need a tool that tracksIn planning a voice over WLAN solution, you the performance throughout the call and is able toshould be considering the network management identify the access point being used during eachand support systems concurrently with the portion of the call.network design. The biggest mistake that ismade in network management is trying to add Identifying/Rectifying Coverage Problems.network management after the network has been Troubleshooting is inherently difficult in a wirelessinstalled. It is absolutely essential that network network, as you cannot “see” the radio signal.management be considered as a critical factor in Anyone can spot a broken wire, but how do youthe overall network design. determine why there is a good signal in one area but not another, particularly when they are both theHere are some of the major areas that should be same distance from the access point? Given theinvestigated: vagaries of indoor radio propagation, there can be vastly different signal readings at points just a fewRF Mapping. Once the network is installed, the feet apart! Training the Help Desk personnel to getnetwork managers should conduct an RF survey accurate location information from wireless usersthat records the signal strength and maps the will be the first step, but many of these problemscoverage area of each access point. That type require dispatching a technician with a test deviceof survey can confirm that the initial design is to the area in an attempt to replicate the problem.sound and is an invaluable tool in troubleshooting When all is said and done, it could just be that thecoverage problems that crop up later. user’s handset is faulty!12 WHITE PAPER: Preparing your WLAN infrastructure for voice
  13. 13. Security. While WPA, WPA2 and 802.1x have Conclusionaddressed the privacy and authenticationconcerns in a wireless LAN, there are other Mobile voice communications over a WLAN cansecurity exposures that will need to be help organizations increase productivity, enhancemonitored. As noted earlier, users or contractors collaboration and, ultimately, improve customerworking in the facility may install unsecured service by making workers instantly accessiblerogue access points on wired network wherever they are in the enterprise. Enabling theseconnections, creating unwanted security enterprise-quality mobile voice services requiresexposures. Attackers may attempt to set up a sound and reliable wireless infrastructure.access points in close proximity to the network in That infrastructure is the result of good planning,hopes of getting client devices to associate with quality tools and a set of systems that will allowthem so they can steal valid user credentials (i.e., the network manager to ensure that the networkuser names and passwords). Also radio jammers is maintained to the highest standards. Tools andor leaky microwave ovens can cause directed expertise are available today that will provide aor accidental denial of service attacks on the wireless LAN voice capability that delivers thewireless infrastructure. The infrastructure must quality and reliability business users expect.include mechanisms to detect, disable and locate With the right tools and a good design plan,these security vulnerabilities quickly. IT departments can deliver a functional and cost-effective mobile voice solution on theirRecord Keeping. You will also have to modify WLAN infrastructure.your ordering and record keeping systems totrack your new class of mobile devices anddefine whether they will be assigned to individualusers or shared by several people within onedepartment.Good-quality voice service requires the abilityto recognize problems before the user calls tocomplain. As these features are not defined inthe standards, it is important to look at whatcapabilities are provided in the WLAN switch andthe handsets to determine what additional toolsand procedures will be needed to help ensure anadequate service level.13 WHITE PAPER: Preparing your WLAN infrastructure for voice
  14. 14. Glossary of Acronyms AC Access Category PRI Primary Rate Interface ACK Acknowledgement QoS Quality of Service Arbitrated Inter-Frame Short Inter-Frame AIFS SIFS Spacing Spacing CAC Call Admission Control System Services SSID Carrier Sense Multiple Identifier CSMA/CA Access with Collision VLAN Virtual LAN Avoidance Voice over Internet VoIP CW Contention Window Protocol/Voice over IP Distributed Control Voice over Wireless DCF VoWLAN Function LAN DCF Inter-Frame Wired Equivalent DIFS WEP Spacing Privacy Direct Sequence WLAN Wireless LAN DSSS Spread Spectrum Wi-Fi Wireless Fidelity Enhanced Distributed WMM Wi-Fi Multi-Media EDCA Control Access Wi-Fi Multi-Media- Information WMM-APSD Automatic Power Save IT Technology Delivery Hybrid Controlled Wi-Fi Protected HCCA WPA Channel Access Access LAN Local Area Network Wi-Fi Protected Orthogonal Frequency WPA2 Access 2 (i.e. 802.11i OFDM Division Multiplexing Compliance) PoE Power over Ethernet WVLAN Wireless Virtual LAN14 WHITE PAPER: Preparing your WLAN infrastructure for voice
  15. 15. motorola.comPart number WP-PYWLAN. Printed in USA 04/09. MOTOROLA and the Stylized M Logo and Symbol and the SymbolLogo are registered in the US Patent & Trademark Office. All other product or service names are the property of theirrespective owners. ©Motorola, Inc. 2009. All rights reserved. For system, product or services availability and specificinformation within your country, please contact your local Motorola office or Business Partner. Specifications aresubject to change without notice.

×