Ericsson RBS 6402 Indoor Picocell
Estimated reading time: 3 minutes, 46 seconds. Contains 756 words
I’m trying to wrap my brain around LAA-LTE. I understand what it’s trying to do, but not why. I mean that makes sense to me. Users are already choosing Wi-Fi over LTE when it's available. LAA will be used for downlink data only and most study's show that Wi-Fi use is significantly greater than mobile data usage anyways. Also, the carriers OWN their own spectrum, and unlike 5GHz, they control it, and have the ability to manage it.
So, why encroach on unlicensed spectrum? They'll have to abide by the power constraints in 5GHz, so in-building, small-cell deployments seem to be the way to go. Does this mean lots of indoor carrier pico-cells right next to existing APs?
I wanted to learn a bit more about LAA and this is what I found so far...
What is LAA? It stands for “Licensed Assisted Access”. Basically, the carriers want to use 5GHz to supplement their “primary” LTE cells with a secondary cell in unlicensed 5GHz. This cell (in the initial stages of LAA-LTE at least) will be for downlink data transmission only. The “License Assisted” part means that the 5GHz cell is linked to the “Licensed” cell via “Carrier Aggregation” (think channel-bonding.. sort of). But, all control functions reside in the licensed spectrum cell and can even disable the 5GHz secondary channel if necessary. At issue here is not the use of 5GHz, which is unlicensed and so by definition anyone can use, but HOW it will be used.
In countries other than United States, China and South Korea, there is a regulatory requirement to “Listen-Before-talk”, or LBT. Since we have no requirement for that in the U.S. carrier manufactures are looking for other ways to place nice with Wi-Fi. The “how” is still up for debate. But, one thing seems clear, they won’t be using CSMA/CA. Qualcomm has a white-paper about this specific issue. Nokia has one to, but you have to buy it. No thanks, I'll just stick with Qualcomm's. Some of their proposed “Coexistence Mechanisms” are:
Channel Selection - attempts to pick the cleanest channel based on Wi-Fi and LTE measurements. Basically, it uses energy detection to see if how used the channel is.
Pretty straight forward, it scans the medium to find the channel that has the least usage. So, in small, to moderately-sized WLAN it should’t be a problem. Especially with DFS enabled.
Carrier-Sensing Adaptive Transmission (CSAT) - "...the cell senses the medium for a duration (around 10s of msec to 200msec) and according to the observed medium activities, the algorithm gates off LTE transmission proportionally."
This one is interesting. If the medium is found to be pretty saturated, and a “clean” channel is not available the cell will fall back to CSAT. LAA-LTE uses an “on-off” duty-cycle pattern. When the medium is heavily used, CSAT will change the duty-cycle timing it will use. It will still use the 5GHz spectrum, but settle on the "least harmful" time-cycle. It also uses only primary channels to mitigate inference with QoS traffic. However, this duty-cycle is directly proportional to the throughput of wi-fi client. In other words, throughout is reduced.
CableLabs.com has a nice piece on this and actually shows the correlation of the LAA-LTE duty-cycle on wireless clients.
Opportunistic SDL - “Since the anchor carrier in license band is always available, the SDL carrier in unlicensed band can be used on an opportunistic base.”
The Supplemental Downlink (SDL) can be enabled, or disabled, on-demand as necessary. In theory, if the primary LTE cell in the licensed spectrum can handle the traffic, the 5GHz secondary cell is turned off. Then if unlicensed offload is required the secondary cell can be turned on. So, LAA-LTE may not always be present. If the carrier's primary channel can handle the load 5GHz won't be used. Question is, will the carriers actually do that?
Qualcomm's white paper claims "due to the coexistence safeguards. In fact, the Wi-Fi performance improves by about 10%, since the neighboring LTE-U Picos can finish transmission faster and incur less interference instead". I don't know about that, and 10% over what?
I'm skeptical about the affect LAA will have on WLANs, but I'm willing to learn and be proven wrong. Also, this is just one more thing we as WLAN professionals have to account for in our wireless designs. 802.11 is already a pretty inefficient protocol and adding more overhead to 5GHz doesn't seem like it will improve anything except the carriers sense of well-being.
Qualcomm White paper:
LTE in Unlicensed Spectrum: Harmonious Coexistence with Wi-Fi
CableLab.com article:
Wi-Fi vs. Duty Cycled LTE: A Balancing Act