Network and Telecom Strategies Blog

Comparing beamforming technologies

posted by: Paul DeBeasi

This is the fifth post in a series that delves into the subject of beamforming.   In this post, we compare all three beamforming methods (static, transmit, dynamic). 

Although all three beamforming technologies provide SNR gain, they are quite different. In addition, all types of beamforming can exacerbate the following problems:

Sticky client problem: The sticky client problem occurs when stations (STAs) remain associated with a distant AP (i.e., when they are “stuck” to an AP), even when a closer AP can offer the STA a better data rate or voice call quality. 

Hidden node problem: The hidden node problem typically occurs when one or more STAs on the same channel cannot “hear” one or more of the other STAs, resulting in channel interference. In some beamforming instances, this can also be true of STAs not hearing the AP.  

A static beamforming array, such as the Xirrus product, is useful in dense deployments such as a large conference room. In addition, a directional antenna can be used to form a static beam to direct energy down a hallway or toward the interior of a building.

The extent to which transmit beamforming (TxBF) will reliably improve SNR is dependent on many factors such as the effectiveness of sounding the channel between the AP and the STA. The WLAN industry is in the early stages of TxBF deployment and I expect that the industry will gain useful real-world experience with Cisco’s proprietary TxBF solution (ClientLink). I also anticipate that silicon vendors will eventually integrate standards-based TxBF mechanisms once the IEEE ratifies implicit and/or explicit beamforming.

Dynamic beamforming is useful in many deployments because it can provide SNR gain for all STA types in the uplink and downlink directions. Since dynamic and transmit beamforming operate in ways that are complimentary, it is conceivable that future enterprise products will combine both techniques to provide even greater SNR gain. Dynamic beamforming will not likely become widely deployed unless WLAN silicon vendors support this feature, thus driving down development cost and complexity.  

Refer to the chart for a more detailed comparison of each beamforming method. 

 Chart: Beamforming comparison

Dynamic beamforming

posted by: Paul DeBeasi

This post continues my discussion of beamforming and in this post we discuss dynamic beamforming. 

Dynamic beamforming focuses RF energy in a particular direction and with a particular shape in order to increase the SNR. In a sense, it is like static beamforming in that it focuses RF energy, but it is different because the antenna array’s radiation pattern can change from frame to frame (see Figure).

 

Figure 1: Dynamic beamforming

Dynamic beamforming is also different from TxBF in several ways because it can:

* Dynamically change the propagation pattern on a frame-by-frame basis in order to optimize the pattern for every STA over time

* Provide SNR gain for both legacy and 802.11n STAs without requiring any changes in the STA

* Be used in conjunction with other 802.11n performance-enhancing techniques such as spatial multiplexing

* Improve both downlink and uplink performance

* Provide interference rejection

Dynamic beamforming has the potential to offer the highest SNR gain of the three beamforming technologies. Note that broadcast traffic, such as Beacon frames, will usually be transmitted using an omnidirectional pattern in order to communicate with nearby STAs in all directions. The only vendor currently offering an enterprise class dynamic beamforming technology is Ruckus Wireless.

Next time we will compare all three beamforming methods. 

Transmit beamforming

posted by: Paul DeBeasi

In this post, we continue my discussion of beamforming by focusing on transmit beamforming. As previously mentioned, beamforming is a method of concentrating radio frequency (RF) energy in order to improve the signal to noise ratio (SNR) at the receiver, thereby improving network performance and predictability.   

Transmit beamforming (TxBF) is a method of transmitting two or more phase-shifted signals so that they will be in-phase at particular points in space where the transmitter believes the receiver to be, thereby increasing SNR. Two forms of TxBF are optional components of the IEEE 802.11n draft amendment to the 802.11 standard.  Explicit and implicit TxBF require feedback from 802.11n stations (STA) and thus will not operate with legacy stations at all. Enterprise WLAN vendors do not support implicit and explicit TxBF at this time because the 802.11n chipsets do not currently support either form of TxBF.

Cisco has introduced a proprietary form of TxBF called ClientLink. ClientLink can work with 802.11b/g/a STAs because it requires no modifications in the STA. ClientLink is designed to improve the SNR for legacy STAs in the downlink (AP-to-STA) direction only. A boost in SNR can improve the STA’s “rate over range” performance because the modulation rate for 802.11 STAs will increase as the SNR increases. Improving rate over range performance is particularly important for legacy STAs because they can consume considerably more airtime than 802.11n STAs, and therefore can reduce the achievable throughput of 802.11n STAs. Alternatively, Air Time Fairness (ATF) mechanisms can also regulate legacy STA airtime consumption.

TxBF changes the phase of the original signals in relation to each other and transmits the phase-shifted signals using two or more antennas to the STA (see figure). As the signals propagate through the air, they additively combine at various points in space.  The figure shows an example of two out-of-phase signals propagating from an AP to a legacy STA. The green dots represent the points in space where the two out-of-phase signals combine to form a signal with an SNR that is up to 3 dB higher than (i.e., twice as high as) the original signal. In a multipath-rich environment, even higher levels of gain are theoretically possible. Cisco is the first enterprise WLAN vendor to implement TxBF, and it claims it can achieve 4 to 6.5 dB of SNR gain in a multipath-rich environment. A 6.5 dB gain is an increase of approximately 4.5 times the original signal.

 

Figure 1: Transmit beamforming

The challenge with TxBF is figuring out how to modify the transmit signal phases for the greatest possible gain. In the ClientLink implementation, the AP uses frames received from the STA in the uplink direction to determine how to modify the phase in the downlink direction. TxBF assumes that the uplink channel characteristics and the downlink channel characteristics are “reciprocal” (i.e., the same) in both directions. In reality, the uplink and downlink channels may not be reciprocal, especially when the STA is moving. So, in practice, TxBF performance gains will vary from moment to moment and STA to STA.

Unlike static and dynamic beamforming, TxBF does not change the antenna radiation pattern. Cisco’s TxBF implementation uses omnidirectional antennas that cause the signals to radiate in a doughnut-shaped pattern. Therefore, referring to TxBF as “beamforming” is somewhat misleading because it does not actually form a directed beam. Cisco’s TxBF achieves an SNR gain at “points in space” (i.e., the green dots in the figure). So the receiving STA must be located at the right point in space in order to achieve the maximum SNR gain. In contrast, static and dynamic beamforming achieve SNR gain throughout the radiated coverage area because both techniques focus the radiated energy.

TxBF and spatial multiplexing are mutually exclusive. This is because spatial multiplexing transmits different signals on each antenna, whereas TxBF transmits the same (phase-shifted) signals on each antenna. So, it is impossible to use TxBF and spatial multiplexing at the same time. That is why enterprises should use TxBF to provide SNR gain for legacy STAs only. In contrast, static and dynamic beamforming can operate in conjunction with spatial multiplexing.

Next time, we will look at dynamic beamforming. 

Static beamforming

posted by: Paul DeBeasi

This post continues my discussion on beamforming.  Beamforming is a method of concentrating radio frequency (RF) energy in order to improve the signal to noise ratio (SNR) at the receiver, thereby improving network performance and predictability.  In this post, we discuss static beamforming. 

Static beamforming provides a fixed radiation pattern by using a directional antenna. Virtually every vendor provides APs with removable antennas so it is easy to swap an OMNI antenna for a directional antenna. Some vendors, such as Xirrus, use an array of static beamforming antennas to create densely packed multiple channels in much the same way that a cellular tower uses directional antennas to create cellular sectors. By assigning a different channel to each beam, many non-overlapping channels can be densely packed together within a single array (see Figure).

 

Figure 1: Static beamforming

Static beamforming can provide an signal-to-noise (SNR) benefit to both legacy wirless protocols like 802.11a/b/g and newer 802.11n stations.  Since the antenna propagation pattern is static, the AP cannot adjust the radiation pattern on a frame-by-frame basis in order to track a station as it moves through the enterprise. Therefore, unlike transmit beamforming and dynamic beamforming, it does not take advantage of knowledge of the WLAN channel between the AP and the station in order to further optimize signal propagation. An array of directional antennas, such as the Xirrus Array, can improve SNR in 360 degrees. 

Next time we'll look at Transmit beamforming. 

Introduction to beamforming

posted by: Paul DeBeasi

Beamforming is a method of concentrating radio frequency (RF) energy in order to improve the signal to noise ratio (SNR) at the receiver, thereby improving network performance and predictability. Enterprise WLAN vendors are now integrating beamforming technology into their access points (APs) so it is important to understand the types of beamforming and the benefits that beamforming can provide. 

Beamforming is not new.  At the most basic level, beamforming affects the radiation pattern of a wireless signal.  The radiation pattern refers to the way in which the electromagnetic waves propagate outward from the antenna element. For example, the most commonly deployed antenna is the omnidirectional antenna. The radiation pattern for the omnidirectional antenna is in the shape of a doughnut (see Figure 1). This type of antenna is a good choice for hotspots or any environment where the intent is to propagate the WLAN signal in a broadly dispersed pattern. 

Figure 1: OMNI antenna radiation pattern

WLAN venders currently offer three types of beamforming.  Each type affects radiation patterns in different ways.

Static beamforming involves the use of internal or external antennas that have a fixed radiation pattern (such as a YAGI antenna) that emits a directional radiation pattern (see Figure 2).  A directional pattern is a good choice in environments where the intent is to propagate the WLAN signal in a particular direction, such as down a hallway or toward the interior of a building.

Figure 2: Directional antenna radiation pattern

Transmit beamforming propagates two or more phase-shifted copies of a signal on a frame-by-frame basis, so that they will be in-phase at particular points in space where the transmitter believes the receiver to be, thereby increasing SNR.

Dynamic beamforming changes the antenna radiation pattern on a frame-by-frame basis using a central processing unit (CPU) controlled antenna array in order to increase the SNR at the receiver.

In subsequent posts we will look at each of the beamforming types.  Next time, we will look at static beamforming.

What to ask before you buy a femtocell

Posted by: Paul DeBeasi

Introduction

Femtocells are all the rage.   These tiny cellular radios promise to improve in-building cellular coverage and are primarily targeted at the home user.  However, some vendors are now building enterprise femtocells.  Should you consider femtocell deployment in order to improve your cellular coverage?  This post identifies the key questions you should ask before you buy a femtocell.

Value proposition

Network operators now offer femtocells such as the AT&T 3G MicroCell, Sprint AIRAVE, and Verizon Wireless Network Extender. The devices provide in-building cellular coverage and look like a Wi-Fi access point.

The devices communicate with the cellular phone and convert voice calls into Voice over IP (VoIP) packets.  The packets are then transmitted over your Ethernet network, through your firewall, and over your broadband connection to the network operator’s servers.

At first glance, the femtocell value proposition seems like a win-win for the customer and the network operator.  The customer receives better in-building coverage.  The network operator not only avoids the cost and hassle of building a macrocell, they also use the customer’s power, internal network, and broadband connection, further saving themselves money.  What’s not to love?

Well, let me count the ways.

Key questions

Before you add femtocells to your telecommunication strategy, you should consider the following questions.

• Which cellular operator’s signal needs improvement?  If you are like most enterprises, your employees probably use cellular service from several network operators.   But a femtocell only supports one operator (e.g., AT&T or Verizon Wireless).  So you may need to install femtocells from many operators throughout your enterprise.

• Where do you need coverage? Oftentimes cellular signals from a macrocell can’t penetrate into the innermost regions of a building (e.g., the basement). Unfortunately, femtocells must be installed near a window so that the attached GPS antenna has a line-of-sight view to orbiting satellites. For instance, AT&T recommends that their microcell be located within three feet of a window and not in the basement or a closet.  So a femtocell might not be able to provide coverage where you need it.

• What type of phone is supported?  The AT&T 3G Microcell supports 3G phones ONLY (sorry, no EDGE support).  The Verizon Wireless Network Extender supports 2G phones but no 3G phones (yup, no EV-DO support).  So it is very likely that many employees will not reap the femtocell benefits because the underlying technology for their phone is not supported.

• How will a femtocell affect battery life?  A femtocell could save battery life because less power is required to transmit a signal over the short distance to the femtocell rather than over the long distance to a macrocell. On the other hand, lots of handoffs from femtocell to femtocell may drain the battery because the phone needs to continually scan for a new femtocell.  So be sure to test phone battery life in your environment.

• How will you identify rogue femtocells? A rogue femtocell is an unauthorized femtocell connected to your network.  In the case of Wi-Fi, vendors provide rogue access point detection software to help detect the rogue AP.  However, what if an intruder configured a rogue femtocell with a valid AT&T account ID (but not the enterprise account ID) and then connected the femtocell to the enterprise network? How would you detect this rogue femtocell?

• How will you isolate femtocell traffic? Femtocell traffic is backhauled across the enterprise network.  Enterprises may want to isolate the femtocell traffic from all other enterprise traffic by using a virtual LAN (VLAN) connection or a separate physical Ethernet connection.

• Is the femtocell solution scalable?  Most femtocells limit the number of simultaneous calls to three or four.  Does this provide sufficient capacity for your enterprise?

• How will you block access to non-employees?  A femtocell signal may spill into an adjacent building or floor.  Most femtocells provide a “white list” or “black list” mechanism to explicitly include/exclude phone numbers.  This is akin to configuring a Wi-Fi AP with MAC addresses for allowed/disallowed laptop connections.  Is this mechanism a manageable solution?  

• How will you manage interference? As with any radio, femtocells are subject to interference.  Unlike Wi-Fi, which uses unlicensed spectrum, femtocells use licensed spectrum that is controlled by the network operator.  So if your femtocell is subjected to interference from another femtocell or macrocell, how will you alleviate the interference?   Note that your ability to move the femtocell may be restricted by the need to maintain the GPS connection.

Conclusion

Femtocells promise to dramatically improve in-building cellular coverage but they raise many questions.  This post identified the key questions you should answer before investing in a femtocell solution.  

iPhone in the Enterprise: Policy drives technology

Posted by: Paul DeBeasi

Enterprises often ask me about the technology solutions that will help them mitigate the risks associated with supporting the iPhone.  Most of the conversations inevitably start with the same questions, discuss the same answers, but, interestingly, lead to different conclusions.   As I contemplate why this occurs, I am struck with the notion that when enterprises ask me about the iPhone they are really asking me to help them figure out their iPhone policies. Policy defines things such as who owns the iPhone (i.e., ownership policy), how the iPhone can be used (i.e., usage policy), and what level of risk they will tolerate by using the iPhone (i.e., risk policy).

Well, if iPhone deployment is a question of policy, then what drives policy?  More often than not, I observe that policy is driven by corporate culture and by the regulatory environment within which the enterprise must operate.  Some enterprises have a very permissive culture. They assume that employees usually try to do “the right thing” and that allowing them to use the iPhone will improve their productivity.  Others, are more restrictive and do not feel comfortable with the lack of enterprise control over the iPhone.  This restrictive perspective is especially true in highly regulated industries such as financial services.

In a “perfect” world, business requirements should drive technology decisions.  But, my observation is that corporate culture has an enormous influence on controversial technology decisions, such as enterprise iPhone support.  Given the fact that there is no technology that will mitigate all iPhone risk, enterprises should spend more time trying to define policies that they can live with, and less time trying to find a perfect technological solution. 

IEEE finally ratifies the 802.11n standard

Posted by: Paul DeBeasi

Does anyone really care that the IEEE finally ratified the 802.11n wireless standard…anyone…anyone…Bueller?

The sorry fact is that the final ratification will have virtually no impact on the wireless industry. This is because what customers care about most is product interoperability. The Wi-Fi Alliance stepped into the standards void in 2007 and began certifying product interoperability based upon IEEE 802.11n draft 2.0. The fact of the matter is that the Wi-Fi Alliance did such a great job with their 802.11n certification program as to make the final IEEE standard a non-event. The evidence for my statement is that the wireless industry currently ships over 1 million Wi-Fi products PER DAY, and well over 50% of those products are based upon 802.11n!

You can relax in knowing that no driver updates (or hardware changes!) are required for compatibility with the 802.11n standard. This is because the Wi-Fi alliance announced that they would not need to make any changes to the baseline requirements of its 802.11n certification program. So, any Wi-Fi CERTIFIED products based upon 802.11n draft 2.0 are automatically certified as interoperable with products based upon the final 802.11n standard.

… I’m still waiting…Bueller?

New WPA attack - updated

Posted by: Paul DeBeasi

Introduction

Back in November of 2008 I wrote about the Wi-Fi Protected Access (WPA) attack by the German graduate students Erik Tews and Martin Beck. They discovered a limited method to crack WPA, or more specifically, to crack the TKIP component of WPA.  Their paper describes the attack and their tkiptun-ng tool carries out the attack.

Now, Japanese researchers Toshihiro Ohigashi and Masakatu Morii have taken the Tews-Beck attack one step further. Their paper describe how they can reduce the attack time from 12-15 minutes down to 1 minute and how they can eliminate the requirement that the station under attack support the IEEE 802.11e QoS mechanism.

Some articles have sensationalized this attack and have even implied that AES-CCMP could be next.

Bottom line:

• The vulnerability exposed by the Ohigashi-Morii attack is no different from the vulnerability exposed by the Tews-Beck. The Ohigashi-Morii attack essentially just speeds up the Tews-Beck attack process.  More specifically, neither the Tews-Beck nor the Ohigashi-Morii attacks discover the encryption key. Instead, they systematically decrypt an individual packet of short length (e.g., ARP packet).
• The attacker cannot decrypt all the packets on a WLAN.
• The attack only exploits TKIP, which is the older “band-aid” feature created to fix WEP (WEP used RC-4 encryption). The attack does not exploit AES-CCMP. 
• If you use certificates on both your stations and your access points (e.g., you use EAP-TLS) then your network is not vulnerable to the Ohigashi-Morii man-in-the-middle attack but your network may be vulnerable to the Tews-Beck attack. 

Recommendations:

• Use AES-CCMP encryption to protect against this attack. All certified WPA2 systems must support AES-CCMP.
• Check to see if your WPA-certified systems support AES-CCMP. If so, you should configure your WPA equipment to use AES-CCMP.
• Begin a program to phase out your older Access Points and stations that do not support AES-CCMP.

Terminology

There is some confusion over the difference between the Wi-Fi Alliance certification and the actual features supported by the vendor equipment.

WPA certification: The older Wi-Fi Alliance WPA certification designation means that the equipment was certified to support: 802.1X, EAP, TKIP, and RC4.

WPA2 certification: The newer (and now mandatory) WPA2 certification designation means that the equipment was certified to support: 802.1X, EAP, and AES-CCMP.

Vendor equipment: Some vendors chose to support both TKIP-RC4 and AES-CCMP irrespective of how the equipment was certified. It is therefore possible to configure some WPA-certified equipment to support AES-CCMP.  It is also possible to configure WPA2-certified equipment to support TKIP-RC4.

More detailed information:

   Erik Tews and Martin Beck paper

Toshihiro Ohigashi and Masakatu Morii paper

Glenn Fleishman article

Maximizing ROI for mobile applications - part 2

posted by: Paul DeBeasi

Introduction

Enterprises use many different approaches to develop applications on mobile devices. This post continues a three part series on maximizing return on investment (ROI) for mobile application development. Part 1 explained why mobile application development is so difficult and introduced three common deployment approaches. This post discusses the benefits and challenges of each approach. Finally, Part 3 will provide specific tips to help enterprises maximize mobile application ROI 

Thin client

The thin client approach uses a mobile browser to access a server resident enterprise application. This approach has the following benefits and challenges.

Benefits

• Application reuse – Many enterprises can re-use their server-based applications by simply providing mobile browser access to those applications.
• Rapid development time – A thin client application requires less development time that the other approaches because the programmer does not need to integrate with different hardware, operating systems, and software development kits.
• Portable – Enterprises can develop thin client applications that are portable across different mobile devices and browsers. One major caveat is that mobile browser compatibility varies from browser to browser (just like on a laptop). The mobile web initiative can help enterprises develop technical best practices to improve user experience on mobile devices.

Challenges

• User unfriendly – Mobile browsers are notoriously difficult to use. Try browsing the web using this opera mini simulator.
• Requires network connectivity – A thin client requires a continuously active network connection to access the application.

Thick client

The thick client approach deploys a full-featured smartphone-resident application. This approach has the following benefits and challenges.

Benefits

• Native look and feel – Programmers develop thick client applications using software development kits (SDKs) that are native to the mobile device. Therefore, the application can take advantage of the these native application programming interfaces (APIs) to achieve a device-appropriate look and feel.
• Hardware access – The developer has access to the mobile device hardware (e.g., GPS, accelerometer) via the native device APIs (e.g., Apple iPhone SDK).

• Integrated security/management – Some platforms (e.g., Windows Mobile and Blackberry) have very strong built-in security and management capabilities.

Challenges

• Limited portability – A thick client is programmed for a specific hardware/software platform (e.g., Nokia E60 and Symbian) and is therefore not easily ported to a different platform. Writing to an application platform (e.g., Java Micro Edition, .NET Framework, Binary Runtime Environment for Wireless) can improve portability but still lacks the portability of the thin client approach.
• Bigger development investment – Any enterprise that wants to support several different types of mobile phones (e.g., BlackBerry, Nokia, and Apple) and uses the thick client approach, must plan for a bigger development and support investment because each platform requires a unique development effort (See www.deviceanywhere.com for testing resources).

Mobile middleware

The mobile middleware approach uses an abstraction layer that sits between the native mobile phone hardware/software and the enterprise application. This approach has the following benefits and challenges.

Benefits

• Portable – Most mobile middleware systems support a broad range of hardware and software platforms. One of the greatest benefits of using a mobile middleware system is the ability to port applications to many environments.
• Integrated security/management – Some platforms (e.g., Sybase iAnywhere Afaria) have very strong built-in security and management capabilities.
• Integrated applications – Some platforms provide built in mobile messaging applications (e.g., email, instant messaging).

Challenges

• Lack of standardization – Mobile middleware APIs are not standardized. Therefore, an enterprise cannot easily change mobile middleware vendors.
• Cost – Unlike many thick client development platforms, mobile middleware solutions are not free.  The enterprise must pay for license fees.

Next time

This post reviewed the benefits and challenges for each approach. In part 3, we look at specific tips to maximize the return on your investment.