**By John Houghton on October 7, 2020** | Leave a Comment

This article was written by John Houghton and proofread for accuracy by Ferris Lipscomb, Ph.D, Solid State Physics, NeoPhotonics.

Let’s say you’re not the technical type and someone in a meeting asks, “If we go from 16 QAM to 64 QAM, how much more bandwidth will we get?” (QAM means Quadrature Amplitude Modulation—a way of packing more data into the same space.) 64 divided by 16 is 4, so we get four times the capacity, right? Not so. In order to get the right answer, you have to think about the symbol rate. Understanding symbol rate was too complex for me at first, but I’ve figured it out with the help of our Ph.Ds so let me show you how symbol rate works. Let’s make it understandable by showing you a table with the answers and then we’ll explain it further below.

The multiplier column above shows how many times faster than On-Off Keying (OOK) as compared to the higher order modulation format. From this you can determine 64 QAM provides 50% more bandwidth than 16 QAM (12 divided by 8). If you’re going from QPSK to 64 QAM you’ll get 300% more bandwidth or 3x. OOK to 64QAM gives you 12x more bandwidth.

**Symbol Encoding Examples**

As I was writing this article, I couldn’t really see how bit rate related to symbol rate without examples, so for those of you who want more detail, let’s go through some examples. Let’s say we wanted to transmit 12 bits, and here they are: 001100110011. Let’s imagine that these 12 bits represent a small graphic and we’ll start by transmitting it with OOK.

OOK – If we transmit 12 bits using OOK, we will use the equivalent space of 12 symbols. OOK is our baseline.

QPSK (also 4 QAM) – If we transmit 12 bits using QPSK, we could do this using 6 symbols. To understand this, let’s dive further. First of all, here is the constellation for QPSK:

When we use QPSK, instead of sending a 0 or 1 we also send data via phase modulation, which gives us 4 different symbol possibilities 00, 01, 10, 11 (versus 2 possibilities). To understand phase modulation, click here. Let’s now encode our 12 bits: 001100110011.

The first two bits are 00. Do you see a 00 in the above constellation diagram? It’s in the lower right. Therefore, instead of transmitting 00, which is two bits, we transmit the whole symbol:

We’ve now encoded the first two bits (00) into a symbol. Let’s encode the next two bits (11). Do you see a symbol that represents 11?

We’ve just encoded the 3^{rd} and 4^{th} bits. You get the idea, so to save time, here is how our 12-bit “string of numbers” looks:

There are 6 symbols. Therefore, using QPSK, we can transmit 12 bits of information in the space of 6 bits by using 6 symbols. Bandwidth is doubled using QPSK versus OOK. But then why does our table say the multiplier for QPSK is 4X and not 2X? It turns out that there is another technique commonly used with coherent encoding techniques like QPSK and QAM, and that is polarization. For more information about polarization, click here. Using these techniques, we can transmit independent sets of data on two separate polarizations, and so we can again double the data rate. So all of the entries for QPSK and QAM in the table have an extra factor of 2 due to polarization. We transmit our 12 bit message using 6 symbols, but divided into two different polarizations which are transmitted at the same time:

Polarization 1:

Polarization 2:

16 QAM has four bits per symbol for each polarization, so to encode our 12 bits, we could break them up into three groups of four, thereby only using 3 symbols in one polarization.

64 QAM has six bits per symbol per polarization, so to encode our 12 bits, we break up the 12-bit string into two groups of six, thereby only using only one symbol for each polarization. We’re transmitting 12 pieces of information, which used to take up 12 symbols, but with the efficiency of 64 QAM, we’re transmitting those 12 pieces of information in one symbols, which take up only one “on-off” cycle.

So back to our original question, if you go from 16 QAM to 64 QAM, you get 50% more bandwidth.

As you can see from the “Total Multiplier” column in the table, the higher QAM you go, the more your returns diminish, at least in terms of the QAM number. For example, going from QPSK to 16QAM gets you 2x more bandwidth, but going from 64 QAM to 128QAM only gets you a 17% improvement, which is not very much. Going from 512 QAM to 1024 QAM only gets you 11%, which is even less. Ignoring polarization, the data capacity increases logarithmically with the QAM number.

**So How Do We Get More Bandwidth?**

Getting more bandwidth from QAM increasingly becomes an uphill battle as we reach farther into the higher orders. Keep in mind also that the greater the QAM the less distance you’ll be able to transmit because higher-order QAM transmission becomes more sensitive to noise.

John Houghton is a Silicon Valley entrepreneur, technology innovator, and head of MobileCast Media.

This article was updated on 10/7/20 to clarify what the addition of polarization means and we did this by adding a column in the table for polarization. Wording was clarified in a number of places by Ferris Lipscomb, Ph.D, and the article was reviewed by other members of our senior technical team.