Advertisement

Chapter 3: Modulation

Using a basic free-space path loss and link-budget calculation, it is possible to determine whether a link is possible. However, that calculation doesn't give you much immediate feedback as to how useful the link will be. The actual throughput of a functional link-budget is determined by two factors: modulation scheme, and channel-width. Additionally, a certain signal quality is required to achieve a given modulation scheme. This section will help you understand channel modulation schemes.

Using a basic free-space path loss and link-budget calculation, it is possible to determine whether a link is possible. However, that calculation doesn't give you much immediate feedback as to how useful the link will be. The actual throughput of a functional link-budget is determined by two factors: modulation scheme, and channel-width. Additionally, a certain signal quality is required to achieve a given modulation scheme – this is the “receive sensitivity” for that modulation scheme.

If radio transmission were always one hundred percent reliable, with absolutely everything sent being received correctly, then we could replace modern modulation schemes with a very simple encoding scheme that layers ones-and-zeroes into the radio waveform and sends them on their merry way, confident that they will arrive in perfect condition. That would let us guaranty that we can use the absolute maximum amount of bandwidth a given channel-width supports for our timing scheme (and obsolete the Shannon formula at the same time!). Unfortunately, the world doesn't work this way: random noise sometimes causes us to lose some data, imperfect aiming, weather, and even birds flying into our signal can all reduce the quality of the received signal. Additionally, we need some way to know whether the data we sent was received correctly. Hence, modern modulation schemes were invented. There are many modulation schemes available, and they all represent a trade-off between the amounts of data sent in a “frame” (time period), versus how much of the frame is dedicated to data that can be used for detecting and correcting errors.

When data is encoded into radio waves for transmission, it is generally encoded against the “phase” of the carrier wave – that is, a sine wave is inverted to represent ones and zeroes. Blocks of data are grouped together, with timing “guard periods” on either side to assist both sender and receiver with correctly identifying where blocks begin and end. In a multi-polarity, MIMO environment (most long-range equipment provides a “2 x 2” link; that is two polarities, and two MIMO channels), there can be four concurrent “bursts” of data at a time. Since it isn't always assured that the data will arrive in one piece, a technology known as “Forward Error Correction” (FEC) is employed: additional data is sent that lets the receiver double-check that the data received matches a checksum in the FEC data, and re-build small amounts of damage if it isn't intact (completely lost frames will still have to be re-sent). Once the burst is decoded, an acknowledgment (“ACK”) is sent back; if no ACK is received, the frame is sent again. Repeated loss of frames can result in unstable links.

There are many different modulation schemes employed in modern Wi-Fi/TDMA links, varying by how many channels they utilize and how much of each burst is dedicated to FEC checksums, as opposed to actual data. The lower the quantity of FEC data required, the faster the overall link can forward useful data. Modern radio systems automatically switch between modulation rates (known as “Modulation and Coding Schemes”, or “MCS”) automatically based upon current transmission success rates. If a link is performing well, then better MCS rates are selected until either a) the maximum MCS is reached, or b) the link starts to degrade, in which case the system reverts to a lower MCS rate. If a link starts to perform badly, progressively lower MCS rates are attempted until one that works is found (or the link dies completely). Additionally, different MCS rates have different requirements for how strong a signal they require to function. For example, on a Rocket M5, the stated sensitivity for MCS15 (the fastest encoding) is listed as -75 dB, while MCS0 (a barely usable data-rate) can operate at -90 dB. The Rocket M5 has the additional restriction that it cannot operate at MCS15 with a transmit-power above 21 dB. (source: http://dl.ubnt.com/rocketM5_DS.pdf )

There are three major encoding schemes used by different MCS levels:

There are sixteen MCS rates supported on Ubiquiti M5 2x2 equipment:

MCS Index Modulation Data Rate (Mbps) 20 Mhz Channel Data Rate (Mbs) 40 Mhz Channel
0 BPSK 1/2 7.2 15.0
1 QPSK 1/2 14.4 30.0
2 QPSK 3/4 21.7 45.0
3 16-QAM 28.9 60.0
4 16-QAM 43.3 90.0
5 64-QAM 57.8 120.0
6 64-QAM 65.0 135.0
7 64-QAM 72.2 150.0
8 BPSK 1/2 14.4 30.0
9 QPSK 1/2 28.9 60.0
10 QPSK 3/4 43.3 90.0
11 16-QAM 57.8 120.0
12 16-QAM 86.7 180.0
13 64-QAM 115.6 240.0
14 64-QAM 130.3 270.0
15 64-QAM 144.4 300.0

The top portion of the table, with the shaded background, represents MCS encodings that can operate with a single polarity (on a single polarity radio, such as the Bullet M5, this is the maximum that can be achieved). The bottom portion of the table lists encoding that utilizes dual polarities (with dual MIMO streams).

Additionally, there are ten MCS rates supported by Ubiquiti M5-AC 2x2 equipment:

MCS Index Modulation Data Rate (Mbps) 20 Mhz Channel Data Rate (Mbs) 40 Mhz Channel
0BPSK 1/2 14.4 30.0
1QPSK 1/2 28.9 60.0
2QPSK 3/4 43.3 90.0
316-QAM 57.8 120.0
416-QAM 86.7 180.0
564-QAM 115.6 240.0
664-QAM 130.3 270.0
764-QAM 144.4 300.0
8 256-QAM173.3 360.0
9 256-QAMN/A 400.0

A common question at this point is: “if I can achieve MCS15, can I really get 300 megabits per second from a Rocket M5?” The answer to that is, sadly, “no”. This is the air-encoding rate, representing the physical amount of data sent over the air-waves in half-duplex. Generally, one has to divide these numbers by between two and three to see actual maximum throughputs – and those can be lowered depending upon the type of traffic being broadcast (many small packets tends to be slower than a few large ones). Additionally, the rates listed are an aggregate of upload and download: if both sides were running at full capacity in a 300 megabit/second link, you would at most see 150 megabit/s traversing in each direction. Finally, the Rocket M5 only comes equipped with a 100 megabit/s Ethernet port – so no matter what you do, it won't exceed 100 megabit/s – in practice it will top-out closer to 95 megabit/s because of Ethernet overhead. Some units, such as the newer NanoBeam/PowerBeam lines and the Rocket M5 Titanium are equipped with gigabit Ethernet ports – and have achieved 150 megabit/s download in real-world testing on near-perfect links.

« Chapter 3: Asymmetric Link Budgets Up To Contents Chapter 3: Combining MCS and Link Budget »

Advertisement