5 ms·

It most certainly did not do that, neither has 5G for that matter. The Shannon's Law refers to https://en.wikipedia.org/wiki/Shannon%E2%80%93Hartley_theorem ht

It most certainly did not do that, neither has 5G for that matter. The Shannon's Law refers to https://en.wikipedia.org/wiki/Shannon%E2%80%93Hartley_theorem https://en.wikipedia.org/wiki/Shannon%E2%80%93Hartley_theore... to see the actual formula the basic principle is that the Capacity C of a comm channel is proportional to the amount of spectrum you have B and the Signal to Interference and Noise ratio (SINR). So basically, if you want more throughput, you need either more spectrum or more cell sites in order to improve the SINR, you cannot simply continue to crank up the power because then you are increasing the co-channel interference (interfering with other terminals on the wireless network). This is why Spectrum is so expensive, the alternative is to build more cell sites, which is expensive. 4G LTE did come a lot closer by using OFDMA but they most certainly did not exceed the limit. 5G using mm-wave also did not, it uses phased array antennas to implement beam forming which created space separation, that is, each user gets their own beam and do not have to share it with others, but within that beam, they still are confined to the law C ~ B * SINR The operators did add a ton of spectrum through various auctions and that plus the new tech is what is driving the tputs up.

tails4e 5y ago
You need more SNR (power), more BW (spectrum) , or more 'channels'. 5G solves this using the last dimension, channels. Beamforming allows the same spectrum to be reused by allowing for spacial separation, i.e. use the same spectrum 5 times by sending 5 beams to 5 physically sperare locations. MIMO is layered on top of this ans uses multipath effects to offer more bandwidth to a single user by taking advantage of the fact each path is its own channel. No violation of the Shannon limit for sure, but very smart and computationally intensive techniques to skirt around the issues of the single isotropic or sector antenna of old.
PaulHoule 5y ago
3G and prior "cellular" systems were inefficient because they can't reuse the same frequency in adjacent cells. See the diagram here https://www.geeksforgeeks.org/frequency-reuse/ https://www.geeksforgeeks.org/frequency-reuse/ Here you have 7 different frequencies used in different groups of cells. In 4G all frequencies can be used in all the cells. If you are between two or three cells you are probably receiving a signal from (and being received by) multiple cells and they are sharing the RF baseband to make it possible. That plus a big bag of tricks let 4G achieve radically better spectral efficiency through "spatial diversity"
- __s 5y ago
  > Here you have 7 different frequencies used in different groups of cells. You should only need 4 different frequencies: https://en.wikipedia.org/wiki/Four_color_theorem https://en.wikipedia.org/wiki/Four_color_theorem For hexagons you can get by with 3, https://en.wikipedia.org/wiki/Hexagonal_tiling#Uniform_colorings https://en.wikipedia.org/wiki/Hexagonal_tiling#Uniform_color...
  - PaulHoule 5y ago
    You'd think so but practically the signal bleeds across to second-adjacent cells and a real channel plan needs more than that. Look at how the 3 WiFi channels in 2.4GHz land is nowhere near sufficient for good spatial diversity and it is even a struggle with the 5GHz channels that aren't gated to avoid interference with radar.
    - __s 5y ago
      Indeed. In fact, the link I responded to shows a pattern matching the 7 color 3-uniform distribution I linked to
  - p_l 5y ago
    That only works if you have exact edges and that each BTW uses only one base channel (iirc, they mix and match as much as possible)
  - zw123456 5y ago
    Yes, each cell interferes with the next, that is call co-channel interference. That is a major factor that limits the capacity of a cellular network. There are a number of approaches used to manage that, carriers spend a lot of money and employ many engineers designing around that. SON (Self Optimizing Network) is the latest tech that is being deployed on 4G and 5G to help combat that. BTW, this is one of the reasons you can get much higher tputs with mm-wave because the beams do not propagate very far which makes keeping co-chan down, that's the good news, the bad news is it doesn't propagate well, which means you need way more nodes.
  - 5y ago
pizza 5y ago
I am out of my element here but I was under the impression that the capacity of an arbitrary wireless network was still an unsolved problem. (Though I presume what you are referring to is a case that is solvable) https://en.wikipedia.org/wiki/List_of_unsolved_problems_in_information_theory https://en.wikipedia.org/wiki/List_of_unsolved_problems_in_i...

tails4e 5y ago

You need more SNR (power), more BW (spectrum) , or more 'channels'. 5G solves this using the last dimension, channels. Beamforming allows the same spectrum to be reused by allowing for spacial separation, i.e. use the same spectrum 5 times by sending 5 beams to 5 physically sperare locations. MIMO is layered on top of this ans uses multipath effects to offer more bandwidth to a single user by taking advantage of the fact each path is its own channel. No violation of the Shannon limit for sure, but very smart and computationally intensive techniques to skirt around the issues of the single isotropic or sector antenna of old.

PaulHoule 5y ago

3G and prior "cellular" systems were inefficient because they can't reuse the same frequency in adjacent cells. See the diagram here https://www.geeksforgeeks.org/frequency-reuse/ https://www.geeksforgeeks.org/frequency-reuse/ Here you have 7 different frequencies used in different groups of cells. In 4G all frequencies can be used in all the cells. If you are between two or three cells you are probably receiving a signal from (and being received by) multiple cells and they are sharing the RF baseband to make it possible. That plus a big bag of tricks let 4G achieve radically better spectral efficiency through "spatial diversity"

__s 5y ago
> Here you have 7 different frequencies used in different groups of cells. You should only need 4 different frequencies: https://en.wikipedia.org/wiki/Four_color_theorem https://en.wikipedia.org/wiki/Four_color_theorem For hexagons you can get by with 3, https://en.wikipedia.org/wiki/Hexagonal_tiling#Uniform_colorings https://en.wikipedia.org/wiki/Hexagonal_tiling#Uniform_color...
- PaulHoule 5y ago
  You'd think so but practically the signal bleeds across to second-adjacent cells and a real channel plan needs more than that. Look at how the 3 WiFi channels in 2.4GHz land is nowhere near sufficient for good spatial diversity and it is even a struggle with the 5GHz channels that aren't gated to avoid interference with radar.
  - __s 5y ago
    Indeed. In fact, the link I responded to shows a pattern matching the 7 color 3-uniform distribution I linked to
- p_l 5y ago
  That only works if you have exact edges and that each BTW uses only one base channel (iirc, they mix and match as much as possible)
- zw123456 5y ago
  Yes, each cell interferes with the next, that is call co-channel interference. That is a major factor that limits the capacity of a cellular network. There are a number of approaches used to manage that, carriers spend a lot of money and employ many engineers designing around that. SON (Self Optimizing Network) is the latest tech that is being deployed on 4G and 5G to help combat that. BTW, this is one of the reasons you can get much higher tputs with mm-wave because the beams do not propagate very far which makes keeping co-chan down, that's the good news, the bad news is it doesn't propagate well, which means you need way more nodes.
- 5y ago

pizza 5y ago

I am out of my element here but I was under the impression that the capacity of an arbitrary wireless network was still an unsolved problem. (Though I presume what you are referring to is a case that is solvable) https://en.wikipedia.org/wiki/List_of_unsolved_problems_in_information_theory https://en.wikipedia.org/wiki/List_of_unsolved_problems_in_i...