For the last 15 years, the booming of data traffic has repeatedly translated into spectrum saturation of dense urban sites. This results in traffic congestion and an impaired quality of service (QoS). Such effects have huge negative impacts on the revenues and image of Mobile Network Operators (MNO’s) who try to differentiate on throughput and QoS.
Spectrum refarming is one of several solutions, each with its pros and cons.
There are generally two ways that a MNO may overcome this saturation issue:
1) Network densification: this consists in deploying new sites in the congested area, hence requires extra-CapEx, and extra- site OpEx, accordingly
2) Improvement of spectral efficiency*, through various possible solutions, e.g.:
o Optimization: several areas of optimization may be envisaged, for example by revisiting the traffic steering strategy, or restricting speech codec rate to half-rate (instead of full-rate) on busy hours. In general, such optimizations are relatively “easy” (as opposed to a network transformation), but may not be sufficient in the longer-term;
o Over-sectorization: a MNO may over-sectorize the sites of concern (e.g. increase the number of sectors from 3 to 4): this is more complex and costly to implement, therefore deployment of a new site is usually preferred to this solution;
o Spectrum refarming: this consists in re-visiting spectrum allocation across Radio Access Technologies (RATs: 2G, 3G, 4G); frequencies previously allocated to legacy RAT’s are re-assigned to new RATs.
What are the principles of spectrum refarming?
Spectrum re-farming allows to …
- free a frequency band from an obsolescent RAT so as to re-allocate it to a new RAT. An example is what is being devised for the European spectrum plans: the 1800 MHz (resp. 2100 MHz) band, historically allocated to 2G (resp. 3G), may be gradually re-assigned to 4G.
- confine this obsolescent RAT in one frequency band just to guarantee coverage for this RAT. Again, looking at the European spectrum plans: 2G (resp. 3G) may be confined to frequencies in the 900 MHz band. This way, any 2G-only customer (resp. 3G Customer) may still access the 2G service (resp. 3G service) through this band
- hence solve QoS issues on sites which were subject to (or close to) congestion.
Eventually, spectrum re-farming translates into the full shutdown of a 3G or 2G commercial service: In Taiwan, after closing the 2G service in 2017, the Regulator imposed that the 3G service be shut down by end 2018, so that, by that time, services will fully rely on 4G. Similarly, AT&T Wireless shut down the 2G service by end 2016.
Allocating spectrum to more efficient RATs allows to process more traffic within the same bands, therefore de-congests dense urban sites and slows down the need for densification (and associated increased CapEx and site OpEx).
A spectrum refarming opportunity should be carefully studied
The feasibility of a refarming project depends on
o the customer base (in terms of its distribution per RAT: 2G-only, 3G, 4G – for subscription and terminal capabilities - and its forecast evolution,
o the traffic per RAT (2G, 3G, 4G) and its forecast evolution,
o the installed capacity per RAT,
o the traffic management strategy across RATs
o QoS objectives
If feasibility is confirmed, the amount of spectrum to be reserved for the legacy RAT would still depend on the above parameters.
There will be financial implications for a MNO to refarm spectrum:
o spending some CapEx for RAN renewal with multi-mode 2G/3G/4G single-RAN BTS,
o spending some CapEx for incentivization on terminals: the amount of CapEx depends on the residual number of legacy RAT subscribers (e.g. 2G-only subscribers),
o accelerating some CapEx spending for the end-to-end provision of new service on the sites of concern: e.g. 4G radios in the refarmed frequency band, possibly fiber-to-the-cell (if this backhauling solution is not already available on-site)
Spectrum refarming is ineluctable. The question is: “when is the best time for refarming?”
The question is not whether or not to refarm the spectrum - spectrum will be refarmed at a certain point in time anyway, as no RAT is eternal - but “when should I do it?”.
When a RAT is in the declining phase (from the traffic perspective), this question should be raised periodically, at least once a year, until the “Go” decision is made:
- the answer is provided through a cost-and-benefits arbitration between at least 2 scenarios (Go and NoGo)
- the “Go” decision is taken based on a full strategic / technical / financial analysis: technical outcomes should be part of the MNO’s Network Master Plan for the coming years.
Last, as in a network life cycle, spectrum re-farming is part of a network transformation (more than just an incremental operation): QoS should be measured after re-farming so as to check whether QoS objectives have been met or not.
* Spectral efficiency (in bit/s/Hz) of a RAT is its ability to transmit a certain throughput in a given amount of spectrum. Spectrum re-farming improves spectral efficiency as new RAT’s are more spectral-efficient than the legacy ones, usually due to breakthrough innovations. Since the launch of GSM in the early 90’s, a number of such innovations have been incorporated in the European Mobile Radio Standards, which allowed to significantly improve spectral efficiency. These include Packet-Switching for Data, W-CDMA access and OFDM modulation (allowing the same frequency to be reused at the same location), higher-rate QAM modulations, turbo-codes, MIMO …