WLAN design basics and wireless network considerations
Wireless is an unbounded medium that does not suffer from restrictions such as wiring. It is flexible and robust and does not have to be protected or routed between specific physical points. For designers, it may be tempting to simply follow the layout of an existing wired network and perform a "cut and paste" for a wireless network. While this process may seem expedient, it will also repeat the wired network’s mistakes and shortcomings and end up costing more in the long-term. As a general rule, less equipment and installation talent are required for implementing a wireless local area network (WLAN) than for a wired network. With wireless, less is often more. Wireless LAN design is a distinct departure from wired design, and the components that make up the framework will be the focus of the next several articles.
How many clients are too many?
Being an unbounded medium, the system design must facilitate efficient propagation of radio frequency (RF, wireless) signals. This facilitates coverage and capacity, which are primary WLAN network parameters. Where does the RF energy reach and how many clients can effectively use it before it becomes over utilized and slow? There are many factors involved in designing a reliable, secure, and available wireless network. Rational planning and design, based on a thorough preliminary evaluation, can ensure that a WLAN will work the first time and allow ease of expansion without disruption. A well-designed WLAN will provide many years of reliable service, and the return on investment (ROI) will be realized much sooner.
One fundamental aspect of WLANs to consider is RF propagation and antenna systems. Using different types of antenna systems to shape the RF "beam" to service specific areas and exclude others can also modify coverage. In most cases, though, an access point (AP) will use an omnidirectional antenna theoretically propagating in all directions. For the sake of this discussion, we will assume this to be a circle with a 200-ft diameter line of sight at an output of 100 mW (20 dBm). This means that client devices within 100-ft of the AP will receive a good signal (> -80 dBm). Obstructions introduce attenuation of the signal by varying degrees depending upon the materials of construction. Plywood and drywall not present much attenuation, while concrete and steel greatly attenuate RF. So does the human body, which is made of more than 60% water. Groups of people can present as much or more attenuation as an elevator shaft made of reinforced concrete.
The upshot is that an AP cannot be simply placed in a convenient location in an office or facility and be expected to provide seamless and error-free network connectivity. The facility or office must be thoroughly evaluated as to suitability for a wireless LAN. This is called a survey and can be either a virtual survey, a predictive survey, or an on-site, physical survey. Unless this is a small office, home office (SOHO) environment, a WLAN design will start off on the wrong foot by neglecting to perform a survey. Some of the items that a survey will determine are:
- The presence of neighboring WLANs
- Other sources of RF that could interfere with the new WLAN
- The propagation pattern(s) of proposed AP(s) required to provide desired coverage
- The effect on propagation due to attenuation by walls and other architectural features
- The effect on propagation due to reflections and other behavior
- The location and type of existing wired network infrastructure, such as wiring closets and copper vs. fiber.
Consider the capacity
A primary issue to be considered is capacity. Capacity is not bandwidth or throughput, but rather the ability of the WLAN to provide reliable and available connectivity to clients in the coverage area. If a given client is in an office served by four APs and everyone is out to lunch, then the owner will have all of the bandwidth and throughput he can use. As people start to drift back in from lunch, the owner may notice the network is slowing down a bit. This is normal in most cases and can be expected; however, what if the network becomes unusable when everyone gets back to his desk and is actively using the network? In all probability, all users within the network are experiencing problems from inadequate network capacity.
This is one illustration of how capacity becomes diminished as client to AP ratios go up. When the design was considered, perhaps the number of eventual clients was not anticipated or predicted. This situation results in more clients using a given AP or APs. As more clients use the AP, there is increased contention for the medium. This requires more transaction time and can result in re-transmissions due to collisions. Only one client can be associated to an AP at any one time. While the transaction may last only a few milliseconds, there are many other clients contending for the use of that AP. Wait times and re-transmissions go up in proportion to the amount of clients contending for the medium.
Capacity is not something that can be accurately calculated in a predictive survey. Sophisticated algorithms will take contention into account, but nothing can predict how people or devices will use the network in real time; mobility of devices complicates this prediction. When a device is mobile, it can roam from one coverage area to another, creating increased contention on the new AP. Quality of service (QoS) algorithms and secure roaming will preempt lower priority traffic on the new AP and could cause further congestion. Mobile clients traveling to the extremes of the coverage area could also experience a loss of contact with other clients, resulting in collisions. Remember that all clients are listening to all transmissions—if a client cannot be heard by anyone but the AP, then other clients will assume the medium is free.
More APs at lower power output
Capacity design comes down to being able to provide adequate bandwidth and throughput for all present and future clients, whether mobile or fixed. This might seem like a tall order and unattainable, and admittedly, not all WLANs provide perfect coverage and capacity. Adding more APs and operating them at lower power output usually solve the capacity problem. This might appear counterintuitive; the key is to bring down the client to AP ratio and in so doing, ensure that using the available bandwidth as efficiently as possible protects that throughput. Lowering the power output avoids interference with other APs and allows for flexibility in architecture; this has the effect of creating smaller coverage areas with a lower client to AP ratio.
Another technique essential for the proper function of WLAN operation and device roaming is using multiple channels for the several APs installed in the facility. Recall that the industrial, scientific and medical (ISM) radio band allows for the use of three nonoverlapping channels, 1, 6, and 11. This allows adequate channel separation and avoids adjacent and co-channel interference. Using a repeating pattern where each successive AP is assigned a different channel, a WLAN can theoretically be extended indefinitely without interference issues cropping up. While a single-channel architecture can be used, it usually requires the use of a central controller to dynamically handle traffic by manipulating power output among APs or by load shedding and leveling; this, of course, adds equipment and operating costs. A multi-channel architecture arrangement allows for a cleaner and more positive transition from one AP coverage area to the next without the use of central management.
– Daniel E. Capano, owner and president, Diversified Technical Services Inc. of Stamford, Conn., is a certified wireless network administrator (CWNA); firstname.lastname@example.org. Edited by Chris Vavra, production editor, CFE Media, Control Engineering, email@example.com.
www.controleng.com/blogs has other wireless tutorials from Capano on the following topics:
WLAN design preparation and needs analysis
IEEE 802.11ah, energy efficiency, extended battery life for wireless devices
Wi-Fi standard designed for large-scale sensor, IoT applications
www.controleng.com/webcasts contains wireless webcasts, some for PDH credit.
Control Engineering has a wireless page.