IEEE Future Directions
Related Articles
Technology Enablers
ICT (Information Communications Technology) has made a significant progress in these last 10 years. 10 years is the usual time it takes for developing a new wireless technology, whose whole life cycle is around 30 years, 10 of development and 20 in operation.
In terms of processing power 10 years means a 64 times increase in capacity. Taking into account that a wireless generation keeps evolving for at least ten more years (till it becomes obsolete by the advance of the new one, we have an evolution roadmap for 4G till 2023) this means that the processing power, in the maturity phase, for 5G devices will be 8,000 time bigger than the one available in 4G devices when 4G hit the market (it might be slightly less than 8,000 times given the slow down of the Moore’s law, but still quite an increase).
What could be the impact of this processing power increase? There could be many, and I want to focus on two of them:
- the dynamic management of the wireless spectrum, and
- the re-balancing of control between the terminal and the network.
Let’s see.
Dynamic spectrum management
In the earlier wireless systems the terminal (the cell phone) could only manage a small spectral “windows” (bands), just one in the very first system; as processing capacity increased so increased the capability to manage more spectral bands and several coding systems in parallel. Present smart phones, as an example, can manage the GSM, GPRS, EDGE, 3G, LTE and WiFi coding systems in several spectral bands and can automatically select the most appropriate one (as an example your smart phones automatically connects, and use, your home WiFi network to save on data wireless cost).
The huge processing power that will be available at the terminal makes it possible to use several bands, from the 700MHz (freed by television broadcast as it has shifted to digital) up to 95GHz and beyond, managing ten and more spectral bands. Here two observations are in order.
The first is that the spectral efficiency (that is the number of “bits” that we can code per single Hz) has now reached its peak with OFMD and 63/256 QAM modulation used in the 4G system. This is the Shannon limit, the boundary beyond which it is not possible to go.
When people talk about increased spectral efficiency in 5G they,well, … cheat! True, we have demonstrators that show an incredible spectral efficiency, current record, I believe, is 145.6 b/s/Hz (with 256 QAM), a miracle result given that the usual spectral efficiency is around 2.5 b/s/Hz.
How is this miracle possible, since it is well beyond the Shannon limit? Cheating, that’s it. The communications takes place using several antennas in parallel, 128 in this case. In this way rather than using a single communication “channel” we use many of them (MIMO: Multiple Input Multiple Output) and the array of antennas coupled with a software that detects and decodes the signal allows the resolution of the interference resulting from multiple channels.
This increased efficiency, hence, is not a 5G property, it is already being used today in WiFi communications (two antennas are normally used) and in the 4G.
The constraints are given first by the available processing power (the more processing power is available, the more parallel channels I can process and therefore solve the interference generated by n-1 channels provided I can have the signals from n antennas), and second by the topology architecture of the antennas array.
The antennas need to be separated one another by at least half a wavelength (that is why you see WiFi antennas shaped like horns, separated by some 10 cm, at the two edges of the box). The higher the frequency, the closer the antennas can be. The use of millimetre waves, 50GHz (at 50GHz the wavelength is 6mm) and beyond, let us create smart antennas array, massive MIMO.
The second observation is that in wireless systems, as in several other areas, one needs to accept a trade off among various parameters /desires. In wireless we basically have that the lower the frequency used the better is the propagation (and less complex the transmitting and receiving equipment), the higher the frequency the more capacity to carry information, therefore the bigger the bandwidth.
The usage of frequencies above 5GHz is critical from the point of view of propagation (easy to imagine how much more critical it would be the usage of frequencies above 50GHz!). The solution is to adopt smaller cells where propagation is less of a concern. Hence, the use of higher frequencies in 5G will multiply the available bandwidth because:
- being the frequency higher it will be possible to use a larger portion of radio spectrum (assigning 10 MHz of band at 1GHz is equivalent to assign 500MHz of band at 50GHz and in 500Mz I can squeeze 50 times more information)
- being the cell smaller there will be fewer users per cell to share the available bandwidth, hence each of them will get more band
- covering the same area will require more cells, hence the available bandwidth over a given area will be multiplied by the number of cells covering it.
