Esteban has been tracking small cell backhaul developments for some time. He pinpoints why and where wireless backhaul will be critical for metrocells, highlights the key factors/requirements driving the choice of technology, and reveals some technical limitations of today's products.
While some people forecast that as many as 80 or 90% of outdoor metrocells will be connected by wireless backhaul, he thinks we'll see something more akin to today's macrocell split – about 55% wireless and 45% fibre. While it's true there isn't fibre everywhere, operators will use that wherever possible and install short range wireless in the vicinity.
So why isn't it going to be mostly wireless backhaul for outdoor metrocells?
In some countries, the availability of fibre is much higher, for example in South Korea, China and Japan. Even in the US, where there is plenty of copper, my guess is that operators will try to reuse existing copper lines to reach the nearest fibre point of presence.
As an analyst, it would be quite easy to be aggressive with high wireless backhaul forecasts, but we have to consider that in reality most operators are very conservative about adopting new technology.
So wireless backhaul may be more important than for macrocells, but not excessively so.
Does that mean locating small cells nearer to these fibre points of presence then?
The most important issue for a Metrocell is to offload traffic from the macro network. Location is fundamental – the NGMN has indicated that cells need to be located within 10 metres of each traffic hotspot – so there really needs to be much more flexibility in backhaul. This is where wireless backhaul becomes more significant.
The role that 3rd party players offering "Small cells as a Service" such as Virgin, COLT etc. can play is also important. The value of street furniture will increase in the coming years. Those who deployed city Wi-Fi in the past (and failed through lack of monetization) have now realised they were getting access to valuable infrastructure. If you get permission for public Wi-Fi and small cell deployment then that could make the Service Provider Wi-Fi business more interesting. This gives intermediate players like Virgin, who can deal directly with building, real estate and council owners, an important role to isolate network operators from many/multiple negotiations with location owners. This would make it much more convenient for operators to contract through intermediate players than directly themselves.
Which wireless technology will be most important?
The issue here is less about technology and more about spectrum. In my opinion, the key requirements are the flexibility to deploy without having to wait for regulatory permission or having to deal with interference from other operators. Taking this into account, the main spectrum choices for mobile operators for small cells are between:
a) Availability of block allocated point to multi-point (P-MP) spectrum. These are microwave frequencies, licensed exclusively in a given geographical area. For example, in Europe 26 or 28GHz or even 42GHz. 42GHz is interesting because it can use a smaller antenna and there is more spectrum available.
One reason why P-MP spectrum is more interesting for metrocells than for macro is that the cells will be positioned at low heights 3-6m above street level. Streets become canyons with good RF isolation between different parallel streets. This leads to much greater spectrum reuse than has been possible before in macro, ultimately allowing the P-MP spectrum to be used more like point-to-point (P-P) because it avoids interference.
Some of the intermediate players may acquire spectrum in these bands. This will be the first place to look for small cell deployment.
b) V-Band (60GHz) is the perfect complement, especially for those who don't own block allocated spectrum. There is plenty of capacity and minimal interference. Spectrum is lightly licensed (effectively free) and the wavelength provides a benefit of small size antennas. This technology perfectly fits small cell backhaul requirements.
By contrast, the E-Band (70/80GHz) requires a larger antenna to meet the regulator requirements of the radiation mask. The larger form factor cannot be quite so easily hidden and integrated into street furniture. Some advances in size reduction are being achieved though, such as E-Band Communications' E-Link Mini.
c) Non-Line-Of-Sight (NLoS). The main challenge here is availability of spectrum, which must be below 6GHz. Some vendors in the industry such as Fastback Networks are promising high capacity also in NLoS conditions, where throughput has traditionally decreased considerably compared to LoS. If capacity is insufficient, then operators can't ensure the quality of experience. So I will be very interested to see what real-world performance these new vendors can achieve and learn how they have surmounted the capacity constraints in sub-6GHz spectrum. Advanced antenna techniques are certainly among the main tools to achieve this.
What about the price of these wireless links?
It depends how you assess the price – whether by cost per link alone, or cost per Gbps throughput.
A typical wireless link probably costs as much as $5 to $10K today. Today, this is particularly challenging for the V and E-Band equipment. The higher the frequency, the more expensive it is. However, the high capacity of these links – as much as 1Gbps – makes the cost comparatively attractive.
But the trend for pricing is down for three reasons:
Firstly, because of volumes – as soon as the number of deployments increase, the benefits of mass production will apply.
Secondly, because of specific silicon chipsets. Earlier, BridgeWave, one of the first players in the 80GHz market, had to rely on discrete components for RF. There wasn't enough market demand to justify specialist RF chipset investment. This increased the cost per unit compared to traditional microwave units, where Broadcom has chipsets available.
The recent announcements of new RF chipsets for 80GHz such as the one from Bridgewave should allow others to reduce their price per unit to that of Siklu which has been using an RF chipset from the outset.
Thirdly, 60GHz has a close synergy with the new Wi-Gig consumer products, which should allow prices to drop even further than for 70/80GHz kit. Thanks to these synergies, 60GHz may get close to the pricing of NLoS, but won't be capable of being cheaper.
Can we expect to see wireless backhaul commonly integrated into metrocells?
The problem with that approach is that there is always some kind of flexibility needed in the backhaul, such as alignment. Since you need to deploy metrocells at a lower height to bring coverage down to street level, in many cases it's going to be quite challenging to integrate both pieces of equipment. I believe there is space for multiple solutions - some will appreciate the integrated box - but in my opinion, the initial market requirement will be for a detached backhaul unit. This doesn't mean that the form factor won't need to take this into account.
So we are already seeing round shape form factors that do not appear to be antennas and which reduce the aesthetic impact when mounted on lampposts. Form factors must be innovative in order not to alert citizens or local authorities that there are two boxes on the lamppost rather than one. There is always public debate around cancer, radiation etc. and it is so difficult to explain the effects of radiation on humans - even when we have more basestations they will be radiating considerably lower power and be much less dangerous overall when compared to inefficient RF signals from macrocells penetrating roofs/buildings etc. Even in these cases, the best solution will be to deploy innovative form factors that don't resemble their macrocell cousins.