In the past I have discussed duplexing modes (FDD and TDD) to see how they affect speed on LTE and 5G. In that article, I touched on the fact that frequency shouldn't affect throughput in theoretical terms, but I never explained why this wasn't the case in the real-world.
It all comes down to penetration. No, not that kind of penetration!
All radio waves come under the category of electromagnetic radiation. Electromagnetic waves travel through materials differently. This depends on a number of characteristics, including the wave frequency.
When a wave passes through a material, it loses a proportion of its power for every metre it travels. This is referred to as the attentuation coefficient. Materials with a high attentuation coefficient are less easily penetrated by waves, such as those from phone masts.
This attenuation coefficient varies dependent on the material density, construction and the wave frequency. Lower frequency waves (e.g. B20/800 MHz) normally travel more easily through materials than higher frequency waves (e.g. B1/2100 MHz) as materials have lower attenuation coefficients for waves at these lower frequencies.
This power loss is most critically an issue with newer 5G mmWave, which uses frequencies upwards of 26 GHz. These frequencies can be blocked using just sheets of paper, let alone brick walls or even just your hands.
Every phone needs to be in receiving and transmitting range of an LTE site to access an LTE network.
LTE on Band 20 (800 MHz) can inherently trancieve (transmit and receive) across greater distances than LTE on Band 40 (2300 MHz), because of the general rule of higher frequencies have higher attenuation coefficients, meaning power loss is higher over longer distances.
It may appear, at first, that this only has an effect on network coverage, but this also ends up having a significant impact on speed too.
On LTE, every device is allocated a number of resource blocks (RBs). These are specific transceiving periods which the network dedicates to the specific UE.
RBs are used for transmitting and receiving data over the network. The number of resource blocks allocated to a UE is dependent on the throughput requested by the UE, as well as what the eNB is able to provide. Often times, the limiting factor for LTE/NR performance is the resource block count available to the UE.
All eNBs have a limited number of resource blocks available to allocate. This total needs to be shared between all UEs connected to the cell site. The total number of RBs available for the eNB to allocate depends on the bandwidth available within its spectrum. You can see the bandwidth to resource block count mapping in the table below.
For a L20 site with 10 MHz bandwidth — this is a typical O2-UK coverage-oriented site — this would be just 50 RBs. Meanwhile, for a site with 3xCA on L1+L7+L20 with 15, 20 and 10 MHz respectively (a typical Vodafone UK capacity site), this is a total of 225 RBs.
Remember from my OFDM article, linked above, that we can calculate the maximum theoretical throughput using the number of resource blocks and the modulation used for data transfer:
We can see that decreasing the number of resource blocks directly affects the theoretical throughput between the UE and eNB. However, remember that those 225 resource blocks must be shared between all devices connected to the eNB, which isn't just yours (unless your're on some super special test site, of course!).
As you modify the frequency, notice how the amount of RBs, and therefore maximum throughput, allocated to each individual UE changes. This is a direct consequence of reaching greater or fewer UEs from one eNB.
While frequency should have no effect on the achieveable speeds in theory, however real-world speeds almost always do relate to frequency.
The extra coverage achieved on lower frequencies normally means that networks deploy a reduced number of sites, resulting in a greater number of UEs connected to a single site. This means that the amount of resources available to individual devices is reduced, meaning that possible speeds are much lower.
If you haven't checked out the previous article, which explores how duplexing modes (TDD and FDD) affect speeds and useability, I'd highly recommend giving it a read.