By now, we have all seen and experienced two inexorable trends in communications: cutting the cord and convergence of voice and data.
Cutting the cord means freedom, productivity, and convenience. Examples include the home cordless phone, the cellular phone, and the wireless laptop.
Convergence of voice and data promises efficiency and lower costs through multi-purpose infrastructure, with voice over IP (VoIP) as the poster child for this promise.
Now these two trends are, themselves, converging in the form of voice over wireless LAN (VoWLAN) a.k.a. wireless VoIP, voice over WiFi, and a few other monikers. The basic idea is to enable the same WLAN to carry both voice and data communications. So whats to help or hinder this convergence? Plenty.
The WLAN, and the IEEE 802.11 specification its based on, was conceived for bursty data communications, which are not real-time in nature and can withstand the instabilities in the wireless link that cause transmission retries, buffering, and other assorted delays, without the user noticing a performance impact. On the other hand, real-time applications, such as voice and streaming applications, are a completely different matter, requiring far greater wireless performance. So, whether you have an existing WLAN or are planning your first deployment, there are some key design considerations that you need to watch for as you proceed.
Top Five Design Considerations for VoWLAN
Building a WLAN to handle the rigors of voice requires close attention to five interrelated design traits.
Compared to data users, voice users will ratchet up the coverage requirement in three ways: location, strength, and uplink stability. Users will naturally expect to hold conversations in locations where data communications are not normally needed the hallway, the stairwell, even the bathroom. While coverage needs to be broadened, it must also be strengthened, so as to ensure toll-grade voice quality. Finally, the uplink (i.e., transmission from the phone to the infrastructure) needs to be far more stable and resistant to variations inherent in the radio signal. Who hasnt had a WiFi connection mysteriously drop temporarily, even while sitting at a desk surfing the Web? Such temporary disruptions will mean dropped calls and very annoyed users.
VoWLAN is attractive to enterprises because it offers the possibility of extending the utility of the same WLAN system to both voice and data users. As one would expect, however, such expanded use will require increasing the overall capacity of the system. This is particularly critical in light of the fact that VoIP transmissions are burdened by high packet overhead, which constrains the number of concurrent voice calls that a channel might otherwise support. The stress of this increased load is further aggravated by the fact that most WiFi phones work only in the 802.11b mode, which uses a lower data rate than the newer 802.11g type, which is found in almost all laptops today.
Once untethered, voice users will be mobile users. And since voice is a real-time application in which packets must be sent at regular and consistent time intervals, it will not tolerate packet processing delays that arise when a phone moves from one access point to another in traditional WLAN systems. So, a focus on seamless mobility will be critical to the success of VoWLAN. More about this later.
Quality of Service (QoS)
By definition, an 802.11 WLAN operates on a shared medium, in which all users must contend for access. Unmanaged contention between voice and data will degrade the system performance for all. In this context, QoS refers to the mechanisms that may be available to address the contention.
Phone Battery Life
It is well known that the overhead of the 802.11 protocol is inherently problematic in causing handsets to consume more battery power than other wireless network protocols, such as cellular.
How the above design considerations will be addressed depends on the WLAN architecture that is chosen for deployment. The market offers essentially two WLAN topologies to choose from.
In a cell planning topology, the available radio channels are distributed among the WLAN access point (AP) (Figure 1). The diagram shows the 802.11b/g case, in which there are only three non-overlapping channels available. Each AP (represented by a hexagon) is assigned a specific radio channel, and then the APs are distributed to form a honeycomb coverage pattern. All the while, the designer must take care to provide sufficient physical separation between any two APs that use the same channel, so as to minimize the interference between them. This is the traditional topology that underpins data-centric WLAN systems.
The channel blanket topology is a recent architectural development. This topology creates extended zones of coverage for every available channel, by using each channel at every AP that is controlled by a WLAN switch. The WLAN switch, in turn, tightly controls the RF channels to prevent the co-channel interference that otherwise plagues traditional WLAN systems. With this approach, radio channels are used to create overlapping channel blankets.
Comparing WLAN Architectures
So, how does each topology handle the five design challenges posed by voice over WiFi operation?
Coverage and Capacity
The objective is to maximize coverage and capacity by deploying APs as close together as possible. Closely spaced APs will ensure the strongest possible signal is received by all users, wherever they are, which makes the wireless link more dependable. Higher AP density also maximizes system capacity, since the closer a user is to an AP, the faster the transmission data rate, and therefore the higher the capacity of each channel.
AP density in a cell planning system is limited by the co-channel interference created by APs that use the same channel. Some vendor solutions vary the transmit power of each AP to try to mitigate the downlink co-channel interference. However, reducing the transmit power may create coverage holes in between the APs.
As the APs are brought closer together, the cell planning system begins to display another phenomenon a drop in overall system capacity. Theoretically, each individual AP is supposed to provide the full bandwidth of its assigned channel. In reality, as two APs with the same channel get closer together, the co-channel interference effect causes them to look more and more like a single cell to the users, thereby actually providing one channels worth of capacity (or less) instead of the theoretical two. The bottom line is that cell planning solutions require trade-off and fine tuning to ensure that capacity, coverage, and, if applicable, variable transmission control are balanced.
The channel blanket topology inherently avoids these tradeoffs. Since the switch avoids, not just reduces, co-channel interference, the system does not have to resort to transmission power control. As a result, APs can be placed as close together as needed to maximize both coverage and capacity. Without the limitations imposed by co-channel interference, deployment becomes dramatically easier, as APs are placed wherever is convenient and desirable, to achieve whatever grade of service the organization wants to deliver to its users.
In terms of capacity, as shown in, the bandwidth of each channel is available to the entire service area that it covers, and the stacking of channel blankets results in three times the capacity, compared to the cell-planning solution for any specific location. Finally, by having every AP receive the same channel, an uplink diversity effect is created, enabling the uplink to be as robust as wireline systems today.
As an aside, consider the importance of a robust connection in the case of another highly-touted trend: WiFi to cellular convergence. This convergence proposes to use dual-mode (WiFi/cellular) handsets that will operate on the companys private WLAN whenever possible, instead of the cellular system, thereby reducing cellular airtime charges. In an ironic twist, early deployments of dual-mode handsets (WiFi/ cellular) have actually resulted in an increase in cellular airtime, in large part because the WLAN systems connection was not robust enough to hold on to the phone and keep it from bouncing to the cellular network as the phone went in and out of WLAN reception.
For voice, the design objective is to minimize or entirely eliminate the delays that are caused by the user roaming between APs.
In a cell planning topology, the user client is associated to one AP at any given time. When the client moves, a handoff from one AP to another must be performed, to enable the client to associate with the next AP. The real-time nature of voice communications demand that sophisticated mechanisms be used to make the handoff as efficient and fast as possible, or otherwise run the risk of experiencing unacceptable delays or even dropped calls as the user moves. Ironically, increasing coverage and capacity through denser AP deployment means that the handoff event occurs more frequently. This stresses the handoff mechanisms in the cell planning solution. In an attempt to mitigate this burden, the industry has been working on a new standard specification, 802.11r, which is to introduce fast roaming.
In contrast, the channel blanket approach eliminates roaming altogether. Thats because the user device actually regards the entire coverage zone of each blanket as a single AP. In such a system, no re-association, handoff of communication or security, occurs. For real-time applications like voice, then, seamless mobility, as opposed to fast roaming, is achieved with zero latency and jitter, ensuring the highest possible quality and reliability of voice communications.
Quality of Service
As stated before, the characteristics of voice and data traffic are such that the two types of communications do not co-exist very well. The objective, then, is to determine how the performance-degrading contention between voice and data will be managed by the selected topology.
The cell planning system inherently requires each AP and each channel to be shared by all contending user types voice versus data, 11b-mode versus 11g-mode, user roles, security levels, etc... To address the issue of voice and data contention, the industry has developed a new standard, 802.11e, which establishes a mechanism for prioritizing voice over data traffic on the same channel. This is a very recent specification, and requires new capabilities in the WLAN user devices, so it is important to determine which device types will support 802.11e and when. Even then, the 802.11e standard will only give a partial answer to this issue, due to the statistical nature of the solution.
The blanket topology can also use the 802.11e mechanism when there are users with data and voice Personal Digital Assistants (PDAs) sharing the same physical channel. The overlapping blanket configuration also enables an additional form of QoS mechanism physical segregation of traffic by channel. In other words, each type of traffic (voice or data, .11g or .11b, high security or low security) can be assigned to a specific channel blanket, so that competing traffic types simply do not compete. This can be done today, with existing clients, with no changes or upgrades required.
Phone Battery Life
By now, it is well known that 802.11 WiFi handsets have dramatically shorter battery life between charges than their cellular counterparts. This is due mostly to the inherent characteristics of the 802.11 protocol. So, while efforts are under way to address this issue via the standards definition, and handset manufacturers strain to develop lower power-consuming chips and better batteries, what can be done on the infrastructure side to make things better today?
The simple answer is: maximize coverage to maximize the data rate at which the device will transmit. The higher the devices data rate, the shorter the transmission time for any given communication. Shorter transmission times will mean lower battery drain and longer battery life. As discussed before, the channel blanket topology will ultimately be more capable in this regard, since APs will be placed as densely as needed, without limitation, to create a solid, uninterrupted blanket of high bandwidth (i.e., high data rate) coverage.
The business case for VoWLAN is compelling, and it may be what finally moves WLAN adoption from limited hot-spot deployment to truly strategic, enterprise-wide use. But achieving the promise will require attention to fundamental design traits of the WLAN infrastructure. As we have seen, it is not wise to assume that WLAN system designed for transactional data communications will automatically support real-time applications as-is this is true for both voice and streaming applications.
Coverage, capacity, and mobility are the three most critical performance dimensions that must be clearly understood and designed for. Quality of service and handset battery life are also important, but will be subordinate and dependant on the first three. Finally, depending on the topology that is selected, different mechanisms may come into play to address these performance criteria (Table 1). IT
David Confalonieri is Vice President, Corporate Marketing at Extricom. For more information on the company please visit www.extricom.com (news - alerts).
If you are interested in purchasing reprints of this article (in either print or PDF format), please visit Reprint Management Services online at www.reprintbuyer.com or contact a representative via e-mail at [email protected] or by phone at 800-290-5460.