Each generation of technology, especially in mobile communications, must be backwards compatible with previous generation, due to market requirements principally. However, in 5G, their will be major differences, in UE and infrastructure, hence, 5G will be totally different, but compatibility with 4G may be available with very limiting constraints in performance (in general).

The main way for low cost development is to be based on software. The 5G infrastructure will contain a very important specs which are virtuality and software defined (like: C-RAN, SDN, SDR...).

Existing 4G infrastructure means macrocells and channel frequencies in the 700 MHz to 2 GHz region, more or less (technically, UHF), and LTE in the radio layer. There's probably nothing to absolutely prevent someone from deploying something they call "5G," with these constraints, but there's also not much improvement you can expect, compared with 4G. It becomes a marketing gimmick, and not much more.

5G is supposed to offer a big increase in the aggregate bandwidth per cell. At least, 10 times greater. Also, greater density of users, in a given area. And while not all of the innovative techniques need to be present in every implemenation, 5G is looking into such new techniques as FBMC instead of OFDM, mm waves instead of the UHF band, massive MIMO instead of MIMO, very large channel bandwidths instead of 20 MHz, and as a consequence of these innovations, very short range in the RF layer, and at least a 10X improvement in the aggregate capacity of each cell, and many more users online, in a given geo location. Also, short latencies, although that has more to do with the max distance to servers than anything else. Still, for 5G to be different, you should expect very small cells, compared with typical 4G macrocells. This is true because more users need to be served per location, at higher bit rates, and because RF range will be reduced, as channel frequencies necessarily have to go way up.

My contention is simple. Marketing types can claim that their new scheme is "5G," even if none of the above innovations are present, but then again, they might as well stay with 4G labeling. After all, 4G can also support small cells, typically for indoor use, but also possible outdoors, higher order MIMO, even if not massive MIMO, and carrier aggregation to 80 MHz, 160 MHz, 320 MHz, and so on. What 4G doesn't do, that makes 5G truly different, is to move way up in the RF frequencies, where truly wide channels are available and where the benefits of massive MIMO, such as beam forming, can be accrued. If you have to stay with the 4G infrastructure, I'm not sure what significant improvements you can expect, even in 4G itself. I mean, 4G is expected to evolve past LTE-A, but even that requires changes to the infrastructure, compared with today's 4G.

In addition to the great information listed above, I would like to add:

Last update from the 3GPP group (responsible for standardizing the 5G), namely Release-15 was all about Non-Stand Alone (NSA) 5G which is the part of the new standard that works concurrently and coexist with the 4G and depend on its infrastructure, more can be found here:

http://www.3gpp.org/release-15

And for the next release (Release-16), there will be more information about the Stand Alone (SA) version of the 5G which is planed to be released during this year (2018).

Hope this helps

5G will be based on FBMC or NOAMA as the basic signal processing techniques.Massive MIMO instead of MIMO.Beamforming optimization for energy efficient system .Green capacities very much close to Shanon. Milimeter wave frequencies with possibility of spectrum ranging from 57 to 63 GHZ.

" 5G will be based on FBMC or NOAMA as the basic signal processing techniques" 3GPP has decided to use OFDM-A for their new Radio.

source: https://5g.co.uk/guides/what-is-5g-new-radio/

Good day to all, in addition to what others said, the 5G network will also reuse the existing IP backbone network to carry cell data to 5G core. A part of the Ip backhaul will be upgrade to carry more and more capacity otherwise how the announced throughput from air interface could be provided without Bauckhaul upgrade. This is my contribution. Thanks