IEEE Future Directions
Related Articles
At the dawn of 4G (LTE) gurus where picturing a rosy scenario. For the first time we had a wireless system that could provide low latency, sufficiently low to make interaction among Things (like cars) viable, and that could be supported by very low cost chips since you could make use of IP in a native mode (no need to go all the way up the ISO stack).
This would have resulted in a widespread connectivity of a variety of things, with an expected 100 billions connected by the end of this decade (Ericsson and Cisco forecast). Word of caution: when you are talking with such big numbers you can basically state whatever number best fit your goals.
- McKinsey forecasted something between 50 and 100 billion connected things by the end of this decade (doesn’t a 50 billion range seems a bit … wide?)
- Gartner forecasted a more precise figure, 20.8 billion (watch out, not 20 nor 22, but 20.8 billion)
- IDC forecasted 212 billion, and
- HP led the pack forecasting 1,000 billion Internet of Things!
The variety of estimates shows that we really don’t know what we are talking about. The basic tenet, anyhow, is agreed by all: more and more things will embed processing and communications capabilities. More and more things will be connected. Here we have the first difference in views depending on the actor, and this is where, I think, we can start to see a difference between 4G and 5G.
The network guys tend to consider as “connected” the things that hook onto the wireless network becoming network termination point (similarly to what a cell phone is). 4G is good for that, since, as we have seen, allows connectivity with native IP. 5G is not different in this respect.
The “other” guys tend to consider as “connected” anything that eventually can communicate its data to something else, independently of where that something else is located, hence making use of networks. A sensors in a garbage bin that measure the “fill level” may be “polled” by public buses as they drive in the vicinity. This let the sensor use very little power (it can even be powered by the vibration generated by vehicles driving by) since it does not need to transmit over a long distance. According to the “network” guys that sensor is not a connected thing, according to the other guys it is. That sensor cannot be connected via 4G (it does not have a continuous radio channel on the 4G frequencies, but it can be connected via 5G since 5G embeds potentially “any” kind of radio signal and radio protocol, included the ones that may be used by public buses to sweep the environment and collect this kind of sensors data. Conceptually 5G goes one layer up in the OSI layer pile, including the session layer, whilst 4G remains tied to the first ¾ layers: physical, link, network, transport. Note that someone would not agree that 4G includes layer 4 since all standardization is focusing on the first 3 layers only for the radio part but to me 4G is a communications system that goes beyond the radio part. You may get different views on this depending on whom you are talking to, but for me this provides a clear idea on one of the differences between the two.
For 5G to embed also the layer 5, session layer, it has to reach out from the network to the edges (in a telco view) and work at the device level. The management of the session has to involve, I would go as far as saying “has to happen” at the device level. This is a crucial difference between 4G and 5G, although seldom emphasized. In the case of the “garbage sensor” the session between the sensor and the monitoring center planning the logistics of the pick up garbage trucks is maintained continuously, even though the physical communications happens only when a public bus sweep the sensor and picks up the data.
This kind of Things will not have a SIM identifying them, for sure not a hardware one. They might have a soft SIM or might even have a completely different identification system.
We can even move a step further and consider virtual things, like software things. A Digital camera can be potentially connected (many already have a wifi or Bluetooth port) and as such fall into the same “garbage sensor” category discussed, but what about individual photo that the camera has taken? My claim is that they could become “things” and could be connected to the web. Today the connection is established via search engines (or social tagging the photo and even the objects, like people, in the photo). Tomorrow we may have that each photo becomes a Thing, and a connected one. As today we can tag a face in a photo on a social network, tomorrow we could have the face becoming an object in the internet space with 5G being the communications fabric.
If you look at IoT in this way it is clear that even the 1,000 billion numbers of HP underestimate the volume.
