Hello to the CWNP Forum.
I have read several posts that include references to Extricom and our "channel blanket" architecture and I have noticed that there has been a discrepency over the understanding of just what our architecture is built and what the net result is when deploying our systems.
The basis of the Extricom system is a design that places all AP's are on the same channel. This we are calling a "channel blanket". You can think of it as a mega-cell. When deploying dual radio AP's (or our new quad radio AP's), the first radio in each AP can be placed on one channel and the second can be placed in the same band on another channel. Variations can include two blankets of G, one in B/G and one in A, etc... Our Four blanket topology gets even more fun and an can dedicate an entire blanket to IDS - full time.
Because the AP's are "ultrathin" they serve no purpose other than the RF communications. A wLAN client associates with the system, as opposed to the AP. As a client roams throughout the coverage area in a blanket, the system makes the decision to route packets to the most appropriate AP for the client in real time, providing zero-latency AP-to-AP handoff. Because all the AP's are on the same channel, the client can roam freely in the "channel blanket" and never know that it is moving from one AP to the next. No smoke and mirrors and no time division multiplexing.
Regarding TX power settings, there are none. Regardless of whether the AP's are placed 5 feet or 500 feet from one another, they operate at full power at all times.
The net result of this is the elimination of co-channel interference which means that you can place the AP's wherever you want to provide complete, not optimal, but complete coverage, and do away with cell planning all together.
The image below is a basic network diagram of a typical Extricom topology.
Now what does this translate to in the real world?
Lets first talk about the term Mobility Zones. A Mobility Zone can be defined as an area of wLAN coverage which exhibits the highest degree of mobile clients, without the clients leaving the specified area of coverage. Translation: a typical floor of a building or half a floor of a building. Therefore, when designing a wLAN deployment, an entire floor, or even multiple floors can be covered by a single set of blankets.
I know I did not cover everything, but I figured that I would give a basic overview and open the floor to Q&A.
So both Meru and Extricom can control the upstream bandwidth of client STA's? Can this be done to each individual STA or just different STA types? What do you consider different STA types? HR-DSSS vs ERP-OFDM?
Love your signature. ;-D
I don't think that Extricom can control upstream flows, but they handle the uplink and downlink processes much differently than Meru. Their APs are essentially just antennae.
Devin, Gene, et al,
I greatly appreciate the time taken to go through the analysis of the Extricom "secret sauce" (I find that a fun term that gets used almost every time we do a presentation/demo).
I'll first address the topic of client collision domain management.
The Extricom system is listening at all times, to all clients, through all radios within a given blanket. This allows for the system to dynamically build a real-time dynamic mapping of all clients and their respective collision domains.
To quote Gene: "When a client STA transmits, it cannot be heard by all of the APs. So, if you are at one end of a building and I am at the other, even though we are on the same channel our transmitters aren't powerful enough for us to hear each other.
Just like cell planning of today, many client devices are on the same channel, but they are in different cells so they can't hear each other."
The diagram below helps to exemplify this concept.
Next I would like to tackle the question on True Reuse:
What is Frequency Re-Use?
By definition it is the multiple use of the same channel, but geographically separated, allowing simultaneous transmission on the same channel, thus increasing capacity without causing interference.
?¡é?€??Natural?¡é?€?? Reuse, however, is rare in practice. Essentially the IEEE 802.11 standard dictates the CSMA/CA ?¡é?€?¡° listen before talk mechanism. The ?¡é?€??Listen?¡é?€?? part is governed by CCA / collision domain with CCA defined as -82dBm, but in reality, NICs are much more sensitive. Therefore, CCA leads to false positives and false negatives which cause delays and lower overall throughput.
The TrueReuse approach is not limited by CCA, providing a dynamic and accurate picture of RF environment, enabling frequency re-use with closer spacing. This leads to the benefits of higher bandwidth for more users, frequency reuse at an increased density, and the avoidance of co-channel interference.
Lastly, the topics of up to 9x aggregate throughput and multiple radios per AP (a point that helps to summarize and bring the two discussion points above into focus).
Lets take the example of an installation using dual radio AP's. I'll use in this example 24 access points, with radio #1 in each AP in 802.11 g (ch 1) and radio #2 in each AP in 802.11 g (ch 11). By using dynamic client collision domain management theory, coupled with TrueReuse in a standard floor plan layout, I can concurrently handle multiple(Tx/Rx) on the same channel in multiple geographical areas of each blanket. Additionally, because I now have two "overlaying" blankets in the same band, using the same True Reuse technology, the system produces 2x times the amount of multiple (Tx/Rx) events. Moving to a 4 radio AP model, using 3 802.11 g blankets and a dedicated 802.11 a blanket for IDS, the system can achieve 3x times the amount of multiple concurrent (Tx/Rx) events. Given standard spacing of AP's, easily achieve a 3x per blanket, giving the system an aggregate of up to 9x throughput.
I go back to our simple diagram below to exemplify this:
I invite any discussions regarding the Extricom system and anyone here to call me at my office line below.
With any technology, or technology variation, there will be a good deal of skepticism, therefore, we are always traveling and performing live demonstrations of our system. What we will bring with us is a switch, three dual radio APs, and some WLAN IP phones. Our setup time is about 5 to 10 minutes (just to unpack the gear) and our demo lasts about 15 minutes for the basic overview.
Questions for mrriccob, (you've opened up a can of "beacons" with this thread ;-)).
Can you point us to a case study of clients that are currently using Extricom in a VOIP enabled environment successfully and not experiencing roaming issues?
How does Extricom plan on handling 802.11n, both legacy and greenfield implementations? Will you be publishing a white paper soon.
How will you perform RTLS?
I see you have responded today to the other questions and this is commendable. You are the "first" in a long time to come out and make your presence known. My "SSID" on to Extricom.
I've tried as much as my simple brain will allow me to follow this post. I don't understand much of it, but it's very interesting stuff.
My point here is to say thank you to Matthew Riccoboni for taking the time to explain Extricom's technology to this audience.
Excellent post Matthew. Thanks for taking the time to do this. I should have a demo kit in the lab soon, so beating this solution up should be fun. ;-) This ought to be an excellent proof-of-concept.
My campus has recently deployed 160+ Extricom APs. One thing that helped me think it through is Layers 3, 2, and 1.
ROU-3-----> ExSW-2------>ExAP-1. The APs are layer one. the APs have no chips, thus no brains. All the thinking, security, and dynamics are taken off the APs and put back into the structure where they belong and have long been established.
Stay tuned for details. Were have deployed a full N (4 Ant) internal and external environment. Very Pleased so far.