Carriers do max out the capacity of many cell towers. Remember that it's you and dozens of other people (at least) concurrently downloading stuff over the same tower. Scarcity is a fact of life, and the bandwidth has to be allocated somehow.
I do wonder how it would work if we did some sort of high-speed dynamic bidding for bandwidth, rather than a constant fee. It could actually be very helpful; it would incentivize carriers to appropriately allocate bandwidth for an area (maximizing profit would coincide nicely with maximizing customer utility) and it would allow customers to directly "feel the pain" of bloated downloads and websites, leading to more market feedback for wasting bandwidth by making 8MB webpages or whatever.
Here's a really cool idea; no SIM card, no network attachment. Your phone uses some technology like Bitcoin micropayment channels (high-speed off-network secure micro payments, great protocol) to automatically buy bandwidth from whichever regional cell tower is offering you the best deal.
The technology is still a bottleneck, but the business model just isn't tightly fitted to the limitations of the bottle neck.
4G cannot viably stream 1080p netflix into the number of homes in America that use it at any given time (unless you use an unrealistic number of cell sites).
Some sort of rationing has to occur otherwise cells would be overrun.
Carriers should bring back free nights and weekends.
Spectra efficiency and even faster data rates are nice but I wonder are carriers going going to upgrade backhaul link and IP core network to allow customers to take full advantage of 5G air interface? This a huge capex spending. I wonder if customers willing to pay even more exorbitant prices for 5G plans for their phone?
With those speeds 5G could be an alternative/replacement for cable/DSL for home internet access. I'm not sure the carriers would be on board with that tough as its its easier to oversubscribe mobile devices than home internet.
>With those speeds 5G could be an alternative/replacement for cable/DSL for home internet access.
Though 3/4G already is, it's what many people use instead of wired internet here (Finland) and it's often marketed as such. I rarely get under 25mbps in our clubroom that only has 4G internet access.
Then it was still in the labs as a prototype. He talks about some of the limitation, the history (3G, 4G, 5G) and how it all set up. It is mostly in the context of building reactive systems and this was just the intro bit.
Also fun fact: since Ericsson controls more than half of the market in smartphone <-> internet gateways and most of those run Erlang. If Erlang stops working, no more cat videos for anyone on their phones.
I hope that the network vendors and the IoT vendors sit down and think about how the devices and the network can work to mitigate things like large-scale IoT DDoS etc. Even if this mean new protocoling or whatever. I'm not sure I'm thinking about this exactly correctly, but I feel there is something in there?
IoT vendors don't think, which is why the crap they sell is vulnerable to so much exploitation.
Network operators should know better than to put cell radios into devices you know will be compromised in a year or two.
Hell, the solution sounds pretty simple to me. Network attached devices must be openly disclosed with open source hardware and must have standardized wired update and control mechanisms. That way consumers can make intelligent decisions to buy products that can see long term support by their communities, because companies absolutely won't do it (and because they are international, you cannot really bind them to support terms of contracts - these negotiations have to happen at the import level, and nothing really stops them from dropping products a year down the road from support and bribing the right people / dissolving and reforming the company to circumvent consequences - but you can force them to disclose source and hardware architecture, enabling the consumers to maintain security themselves).
Cellular IoT doesn't really have the problem seen with Mirai & co DDoS attacks. A cellular device is attached to a cellular network first, and for many application are not even on the Internet.
In 3GPP cellular IoT (2G/GPRS, 3G/UMTS to HSPA, 4G/LTE) it is very common for devices to be on what's called a private APN. It's a private IP network, and the cellular operator provides a VPN access to this private network to let the managing company access the devices from their application servers. Only the application servers can access the devices, which are not on the Internet. In practice this removes the risk of remote attacks on the devices (if you either hacked the telco or service operators and can get to the devices, you already could get to more interesting stuff like aggregated data). This has been available since GPRS and will also apply to 5G --- no difference at this level.
Of course the flip side of this model is that the end customer must go through the device associated service provider to get to the device. It may not be acceptable to all, but when it is using a private APN is really a very effective way to kill remote attacks on cellular IoT devices.
Even for devices not using a private APN but having Internet access, the operator is still in between and could do ad-hoc filtering if called for. Another factor is that tier 1 telco operator tend to require support for FOTA (Firmware update Over The Air), so cellular IoT devices can be remotely upgraded.
With all this cellular IoT is in a much better situation for security than LAN-based IoT. It's also more expansive.
I'm really interested in the potential for p2p networking in 5G. But I've been disappointed that LTE Direct[1] hasn't really become a normal thing since a couple years ago when Qualcomm announced it. A cursory googling of 5G shows that some amount of 5G research has focused around device-to-device, mostly for IoT purposes[2][3]. Link [3] in particular talks about the challenge of incentivizes:
> Besides the technical challenges, there is the very practical problem of incentivizing users to lend their devices to serve as relays for the traffic of others, especially since these connections will consume bandwidth, storage, and battery power on the relay. Again, the appropriate class of solutions depends upon the D2D design.
If this D2D does become a central part of 5G, I'm really curious about whether users will be able to use 5G not (only) with a carrier's towers, but as a part of a mesh network (and as a Bluetooth- or wifi-like tool for local networking). It could be a great way to chip away at the power of carriers and ISPs (which might be one reason LTE Direct is nowhere to be seen, and why D2D 5G capabilities could be gimped in end-user devices).
I'm a complete layman in this area, so I'm curious to hear others' thoughts.
As an end-user, I'm ok with current 4G/LTE link speeds. What would be great is they could minimize TTFB on mobile devices. It takes a second or two for the first bits of content to appear in the browser/mail, especially after the phone has been in sleep mode.
The only reason the speeds seem fine is because due to data caps people are too afraid to actually use the data. With more capacity in the networks, data caps can rise, and usage will go up. I don't want to have to mess around with connecting to coffee shop WiFi networks all the time. Every computer should have a wireless chipset that simply uses the WWAN when a WLAN is not available, without you worrying about that OS update downloading in the background.
When I went to the US two years ago I got LTE in the Baltimire, Washington DC and Los Angeles areas so you guys do have true 4G. The issue is more the amount of LTE coverage avaliable which isn't that great outside of the main cities (or at least it wasn't at the time I visited).
Does anyone else restrict their phones to 3G because 4G (real 4G) eats through your data cap too fast?
If I accidentally load a large video/gif/website, It's easier to back out then let it eat my data.
I'm also noticing that with similar usage, my same apps are eating much more data than they used to a couple years ago. Twitter, Facebook, Maps. I could live on 500MB/mo, now 2GB/mo is not enough.
Yes. The issue is mitigated by massive beam-forming. Thanks to the high frequency antennas are small and it's possible to use many of them to steer a narrow beam toward the device. The antenna gain compensates for the pretty tough link budget at those high frequencies.
For the number of antennas, I've seen anything from 64 to 1024 (amazing...). Even then millimeter waves are only for short distances (< ~300m).
Of course, a very narrow beam makes life very interesting with mobile devices ;) But one of the first application of mmWaves should be last mile access, where this is not an issue. With this, fiber is required close to the home but the most costly last hundred meters can use wireless. This should reduce cost while still supporting very high throughput. TBC.
[+] [-] callesgg|9 years ago|reply
I would prefer if we aimed our research on increasing coverage.
[+] [-] the_mitsuhiko|9 years ago|reply
5G is for more countries than just the US. Plenty of countries are very happy with the local carriers.
[+] [-] wyager|9 years ago|reply
I do wonder how it would work if we did some sort of high-speed dynamic bidding for bandwidth, rather than a constant fee. It could actually be very helpful; it would incentivize carriers to appropriately allocate bandwidth for an area (maximizing profit would coincide nicely with maximizing customer utility) and it would allow customers to directly "feel the pain" of bloated downloads and websites, leading to more market feedback for wasting bandwidth by making 8MB webpages or whatever.
Here's a really cool idea; no SIM card, no network attachment. Your phone uses some technology like Bitcoin micropayment channels (high-speed off-network secure micro payments, great protocol) to automatically buy bandwidth from whichever regional cell tower is offering you the best deal.
[+] [-] rhino369|9 years ago|reply
4G cannot viably stream 1080p netflix into the number of homes in America that use it at any given time (unless you use an unrealistic number of cell sites).
Some sort of rationing has to occur otherwise cells would be overrun.
Carriers should bring back free nights and weekends.
[+] [-] artursapek|9 years ago|reply
[+] [-] bogomipz|9 years ago|reply
With those speeds 5G could be an alternative/replacement for cable/DSL for home internet access. I'm not sure the carriers would be on board with that tough as its its easier to oversubscribe mobile devices than home internet.
[+] [-] endless1234|9 years ago|reply
Though 3/4G already is, it's what many people use instead of wired internet here (Finland) and it's often marketed as such. I rarely get under 25mbps in our clubroom that only has 4G internet access.
[+] [-] pault|9 years ago|reply
Edit: My hotspot tends to run at around 22Mb/sec, which is faster than the fastest DSL line you can get here, which is 16Mb and costs $400/mo.
[+] [-] Thaxll|9 years ago|reply
[+] [-] muterad_murilax|9 years ago|reply
Should probably be Kista:
https://en.wikipedia.org/wiki/Kista
[+] [-] rdtsc|9 years ago|reply
https://youtu.be/rQIE22e0cW8?t=603
Then it was still in the labs as a prototype. He talks about some of the limitation, the history (3G, 4G, 5G) and how it all set up. It is mostly in the context of building reactive systems and this was just the intro bit.
Also fun fact: since Ericsson controls more than half of the market in smartphone <-> internet gateways and most of those run Erlang. If Erlang stops working, no more cat videos for anyone on their phones.
[+] [-] neom|9 years ago|reply
[+] [-] zanny|9 years ago|reply
Network operators should know better than to put cell radios into devices you know will be compromised in a year or two.
Hell, the solution sounds pretty simple to me. Network attached devices must be openly disclosed with open source hardware and must have standardized wired update and control mechanisms. That way consumers can make intelligent decisions to buy products that can see long term support by their communities, because companies absolutely won't do it (and because they are international, you cannot really bind them to support terms of contracts - these negotiations have to happen at the import level, and nothing really stops them from dropping products a year down the road from support and bribing the right people / dissolving and reforming the company to circumvent consequences - but you can force them to disclose source and hardware architecture, enabling the consumers to maintain security themselves).
[+] [-] yaantc|9 years ago|reply
In 3GPP cellular IoT (2G/GPRS, 3G/UMTS to HSPA, 4G/LTE) it is very common for devices to be on what's called a private APN. It's a private IP network, and the cellular operator provides a VPN access to this private network to let the managing company access the devices from their application servers. Only the application servers can access the devices, which are not on the Internet. In practice this removes the risk of remote attacks on the devices (if you either hacked the telco or service operators and can get to the devices, you already could get to more interesting stuff like aggregated data). This has been available since GPRS and will also apply to 5G --- no difference at this level.
Of course the flip side of this model is that the end customer must go through the device associated service provider to get to the device. It may not be acceptable to all, but when it is using a private APN is really a very effective way to kill remote attacks on cellular IoT devices.
Even for devices not using a private APN but having Internet access, the operator is still in between and could do ad-hoc filtering if called for. Another factor is that tier 1 telco operator tend to require support for FOTA (Firmware update Over The Air), so cellular IoT devices can be remotely upgraded.
With all this cellular IoT is in a much better situation for security than LAN-based IoT. It's also more expansive.
[+] [-] johnchristopher|9 years ago|reply
[+] [-] nathcd|9 years ago|reply
> Besides the technical challenges, there is the very practical problem of incentivizing users to lend their devices to serve as relays for the traffic of others, especially since these connections will consume bandwidth, storage, and battery power on the relay. Again, the appropriate class of solutions depends upon the D2D design.
If this D2D does become a central part of 5G, I'm really curious about whether users will be able to use 5G not (only) with a carrier's towers, but as a part of a mesh network (and as a Bluetooth- or wifi-like tool for local networking). It could be a great way to chip away at the power of carriers and ISPs (which might be one reason LTE Direct is nowhere to be seen, and why D2D 5G capabilities could be gimped in end-user devices).
I'm a complete layman in this area, so I'm curious to hear others' thoughts.
[1] https://www.qualcomm.com/invention/research/projects/lte-dir...
[2] https://www.ericsson.com/research-blog/5g/device-device-comm...
[3] http://www.comsoc.org/ctn/device-device-communication-5g-cel...
[+] [-] rodionos|9 years ago|reply
[+] [-] kalleboo|9 years ago|reply
[+] [-] nickbail3y|9 years ago|reply
[+] [-] kiwijamo|9 years ago|reply
[+] [-] fudged71|9 years ago|reply
If I accidentally load a large video/gif/website, It's easier to back out then let it eat my data.
I'm also noticing that with similar usage, my same apps are eating much more data than they used to a couple years ago. Twitter, Facebook, Maps. I could live on 500MB/mo, now 2GB/mo is not enough.
[+] [-] petra|9 years ago|reply
http://www.phonearena.com/news/Bandwidth-Ruler-for-Android-l...
[+] [-] djrogers|9 years ago|reply
Wow, 15Ghz signals are going to really suck for building penetration...
[+] [-] yaantc|9 years ago|reply
For the number of antennas, I've seen anything from 64 to 1024 (amazing...). Even then millimeter waves are only for short distances (< ~300m).
Of course, a very narrow beam makes life very interesting with mobile devices ;) But one of the first application of mmWaves should be last mile access, where this is not an issue. With this, fiber is required close to the home but the most costly last hundred meters can use wireless. This should reduce cost while still supporting very high throughput. TBC.
[+] [-] CodeSheikh|9 years ago|reply
[+] [-] kiwijamo|9 years ago|reply
[+] [-] ryanmarsh|9 years ago|reply
[+] [-] sorenjan|9 years ago|reply
[+] [-] mtgx|9 years ago|reply
[+] [-] kristoffer|9 years ago|reply