Tuesday, October 26, 2010

More Power Means Less

While assembling some notes from a recent working group meeting, i was reminded of a discussion regarding Wi-Fi client power management; or lack thereof.  This is a major factor in maximizing available capacity for Multimedia-Grade environments given the density and close proximity of APs in such deployments. The problem, simply stated, is that a Wi-Fi client's RF power is set manually, not negotiated with the AP as is common to cell phones and their base stations.  For many clients, I've seen this power level set to max most of the time.  When the client's radiation, be it on-channel or adjacent channel, reaches another AP to which it is not associated, it can cause interference to that APs channel, degrading its throughput.  I've seen cases where in a "non-optimized" environment, the interference is so bad, it shuts the AP down.  

Fortunately help is on the way.  The Working Group for 802.11k is developing what are called "Cooperative Control" mechanisms between the AP and the client.  These are direct and positively acknowledged conversations that among other things, will allow the AP and client to negotiate power levels to optimize performance while minimizing interference.  Here is a mummery of relevant 802.11k features:
  • Beacon Report - The client reports Beacons that it detects to the AP.  This gives the AP and its associated controller more information about the environment seen by the client.
  • Neighbor Report - The AP sends a list of neighbor APs to the client.  This is used to reduce scanning.
  • Power Constraint Element - The AP instructs the client device to change (typically reduce) its transmit power.
  • Link Measurement Request/Report - The AP can ask the client o report the link quality it is seeing.

To learn more about the power of "Cooperative Control",  check out the 802.11 webpage:

Wednesday, October 20, 2010

The Multicast Option

In the quest to preserve precious wireless bandwidth, multicast can be a beautiful thing; especially for video if you have the right paradigm.  Anytime video content is destined for the masses and is streamed live or its delivery can be scheduled a la the broadcast television model, multicast can save you bandwidth big time.  For example, if I were to stream CNN live, I could deliver standard (not HD) broadcast quality using MPEG-2 at 2Mbps.  If I have 50 people associated with an AP and 20 of them chose to watch CNN, a unicast delivery scenario would mean each user would receive their own copy of the broadcast.  That's 20 users times 2Mbps or 40Mbps of wireless bandwidth being steadily consumed.  If you’re running 802.11n, you'll probably be able to accommodate the load but at the expense of available bandwidth for other users.  Running 802.11b/g/a means you're dead in the water without off-loading that traffic to other APs.  In a multicast scenario, the content is streamed only once and every user accesses that same stream.  Thus only 2Mbps is consumed leaving plenty of bandwidth remaining.

Seems like an ideal scenario doesn't it?  However, multicast suffers from one key disadvantage. Multicast traffic is not acknowledged. Clients cannot indicate to the transmitter that they missed a packet, and because there is no retransmission mechanism, any errors due to lost packets cannot be corrected.  Multicast over Wi-Fi compounds this difficulty. Wireless frames are subject to loss and corruption over the air.  These factors are addressed in a unicast connection using 802.11 protocol features such as acknowledgements, retransmission and rate adaptation.  But with 802.11 multicast, there is no acknowledgement or adaptation, and therefore some level of frame loss is inevitable.  The error rate can be reduced by adjusting the modulation rate, given a constant over-the-air signal to noise ratio (SNR). For instance, if the rate is reduced from 48 Mbps to 24 Mbps, the error rate will be improved provided that the noise level does not change. Because of the lack of acknowledgements, 802.11multicast traffic is usually transmitted at a much lower rate than would be used for unicast traffic. This takes more time on the air, consuming more of the network’s data capacity, but provides a margin of safety in case RF conditions deteriorate.

An additional challenge with multicast over Wi-Fi is that the modulation rate is set for the worst-case among the client population; normally the client most distant from the access point. For example, if four clients on an access point subscribe to a multicast group, and they would connect with unicast traffic at 36, 36, 24, and 18 Mbps, then the multicast stream must be transmitted at a maximum of 18 Mbps. As noted above, a safer figure would be 12 or 9 Mbps giving a better SNR to improve error rates and thus throughput.

So, while multicast can preserve wireless bandwidth, the reality is it may take some away due to increased errors rates.  To effectively propagate the content, the modulation rate must be lowered at the expense of network capacity.  Fortunately, wireless infrastructure is becoming more flexible and accommodating allowing you to select the best methodology for multimedia propagation.  Here are some of the key multicast optimization techniques for supporting multimedia in a campus environment:
  • The infrastructure should automatically adapt by keeping track of the transmit rates sustainable for each associated client and using the highest possible common rate for multicast transmissions.
  • Wireless needs to support IGMP snooping and IGMP proxy, ideally at the central controller, so that it can identify which APs and clients need particular transmissions, blocking all others. This adds significant efficiency to the overall network.
  • The network should automatically select the best transmission mechanism based on real-time network and video usage information. When multicast is transmitted as unicast over the air, it can be transmitted at much higher speeds and has an acknowledgement mechanism to ensure reliability. The network should make this conversion when appropriate and then automatically switch back to multicast when the client count increases high enough that the efficiency of unicast is lost.


Monday, October 18, 2010

Scaling All Things Mobile

I was on the road last week week to Educause.  For those of you not aware, Educause is THE premier conference for IT in Higher Education.  Folks were buzzing with tales and expressing keen interest in all things mobile with the iPad, iPhone and Android being front and center.  Engaging in numerous conversations myself, the central theme emerging from those discussions was all about scaling wireless access to accommodate the onslaught of mobile devices while satisfying their thirst for multimedia bandwidth.  It was interesting to note that many users now carry more than one device; typically a converged, mobile phone plus an iPad and/or laptop.  Also intriguing was that for many, these mobile devices aren't spread out across their environment but clustered into areas with high population densities.  As if connecting the sheer numbers of devices isn't enough of a challenge, accommodating highly dense clusters of these multimedia "bad boys" increases that challenge by an order of magnitude.

Prior to my trip, I had the opportunity to review a document focused exactly at addressing these challenges.  Aruba Networks produces a series of documentation known as Validated Reference Designs or VRD's.  These are templates aimed at giving you deployable examples of wireless designs to meet various criteria.  But the real value I've found in VRD's is that their templates have been built and empirically tested with test criteria and results fully documented.  Aruba has just released the latest in their series; the High Density VRD.  The HD VRD explores numerous techniques for scaling bandwidth in highly dense, high capacity environments using RF spectrum management coupled with selective AP placement strategies.  This includes orthogonal channel selection methodologies with channel re-use, band-steering and AP/client power management just to name a few.  Most impressive was the amount of research given to AP placement including antenna placement and polarization strategies that effectively use elements of the environment in which they're deployed.

I found the HD VRD to be a most informative and enlightening read being directly applicable to the scaling challenges I discussed with my Educause colleagues.  I encourage you to read it for yourself.
http://www.arubanetworks.com/technology/design_guides.php

By the way, Educause is at the Anaheim Convention Center this year (i.e. Disneyland).  I was able to catch their free outdoor Wi-Fi and FaceTime Mickey to the kids as I strolled the streets of Downtown Disney.

It truly is a highly mobile, multimedia world!

Monday, October 4, 2010

More Space For Your Wi-Fi Place

One of the biggest challenges in scaling a network to Multimedia-Grade is delivering the capacity required to support those bandwidth hungry applications and content.  With the proliferation of multimedia-savvy mobile devices, bandwidth consumption is not only going to increase but it's demand will be ubiquitous because of their mobility.  If we're to meet that demand, we must plan for scaling capacity uniformly across the network.  Fortunately, many methodologies exist that can be used to our advantage to mitigate the challenge.  For example, you could deploy additional channels over which to spread the load.  With careful planning of the Access Point's (AP) physical deployment and assistance from the controller for channel management, channel re-use becomes viable, especially at 5GHz.    You could also increase the number of AP's within the same geography, letting the controller load-balance the traffic as required to increase the available bandwidth within that geography.  Moving from 801.11a,b,g to 802.11n is a clear win towards dramatically increasing capacity though support for legacy devices may hinder it's effectiveness in the short term.

These examples will sustain us for awhile but, ultimately, will only take us so far.  The RF spectrum supporting our networks is clearly a limited resource.  At some point, we're going to need more spectrum.  In fact, a lot more of it if we're to keep pace with the unbridled growth of multimedia for mobile devices.  There was a recent article in PC World discussing a novel concept to utilize White Space.  White Space is the RF spectrum reserved between each TV channel to minimize the interference between those channels.  Turns out this White Space contains plenty of room to support wireless data in addition to its mission of preventing interference between TV channels.  It's a novel concept that could provide some relief for our spectrum needs without affecting its current users.  Not affecting those users could dramatically reduce the amount of time to win an allocation making more spectrum available sooner than later.

Check it out!