How remotely controlled antennas and Artificial Intelligence will transform mobile networks
Mobile radio networks have seen a tremendous growth in traffic since the first 2G networks were deployed 30 years ago. While it is tempting to believe that the capacity boost is a result of more effective radio protocols (3G, 4G and soon 5G physical layers) and more spectrum, the actual main contributor is ”higher spectral efficiency”.
To understand what this means in English, consider a large room with only two people speaking to each other. Then add another 100 people to the same room. These people gather in pairs and start talking to each other simultaneously. The amount of information being exchanged in the room is now increased by a factor 100. As everyone is using the same, mechanical airwaves (spectrum), the spectral efficiency is increased by the same rate.
If we apply this to mobile networks, densification of base stations, which translates to reuse of spectrum, has contributed to about 99% of the total capacity increase over the past decades (referred to as Coopers Law). Just as the 100 people in the room can not stand too close to each other for this to work, radio base stations are separated from each other geographically. Inter-site distance between two base stations in a city can sometimes be less than one hundred meters. With the advent of 5G systems, more and smaller cells will be deployed both indoors and outdoors.
Energy sent out from a base station is directed by antennas in the same way that you steer the energy in your voice by the direction of your head. If we take the analogy with the room filled with people again, you typically want to point your mouth in the direction of the listeners’ ear. The same goes for mobile base station antennas which today are vertically tilted and horizontally directed to serve targeted areas and not interfere with each other. The alignment can be done manually, digitally (Massive Multiple Input Multiple Output, mMIMO) or with small engines that adjust the vertical tilt. If the tilt can be adjusted remotely by small motors, we are talking about Remote Electrical Tilt (RET).
RET antennas are cost-effective compared to massive MIMO antennas, but normally the tilt is only set once at installation. The reason for this is that operators are not willing to jeopardize the tuning of their networks once things are working. Changing the tilt of a single antenna could result in unwanted side-effects (dropped calls or increased interference) in other parts of the network, and it takes experienced network optimizers to figure this out.
Given the quick evolution of machine learning and artificial intelligence, we can expect that the intelligence behind adaptive antenna tuning is handled by software in the near future, and then the value of reliable RET antennas will grow rapidly. Consider an example where an antenna could cover an industry area during daytime, a highway in the afternoon and a residential area in the evening. AI software would figure out and tilt all antennas in the area in order to optimize overall system resources.
One common challenge with RET antennas is that they tend to break over time. A lot of issues are related to the signaling cables and how those are standardized, but another important aspect is that they are built with stepping motors which are prone to break if dust finds its way into the bearings. They are also built with parts that can rust. By using linear motors instead, the lifetime and reliability of the RET antennas can be vastly improved.
Why motor controlled filters are needed in 5G base stations
Over the past decades, we have got used to connecting our smartphones, cars, tablets and wearables to mobile networks. On top of this infrastructure, services like WhatsApp, Netflix, Uber and Spotify have transformed industries.
The foundation of this evolution is the advance of mobile network infrastructure (2G, 3G and 4G). Up til now, three to five networks have been built in parallell in each country by Mobile Network Operators (MNOs). This was for long a working model, but as the Added Revenue Per User (ARPU) is declining every year, necessary network densification is folded by many MNOs.
At the same time, the next generation of networks (commonly referred to as ”5G”) are about to be rolled out. When in operation, 5G will be more cost-effective to operate than earlier mobile networks, but the cost of rolling out new equipment and buying more spectrum is substantial.
The best way for MNOs to cut costs is to start cooperating with each other. Theoretically they could build a single physical infrastructure per country and then share the Capital Expenditures (CAPEX), Operating Expenses (OPEX) and license costs. For legacy reasons, they are not very interested in doing this, but are in face already sharing infrastructure in many countries, a trend that will continue. In Germany the MNOs have expressed a will to build out a single, common infrastructure for 5G in rural areas.
When operators are sharing radio equipment, it is favorable if those are built with multi-standard (2G, 3G, 4G, 5G) and multi-frequency components so that spectrum and access technologies can be added or removed adaptively and remotely. This saves a lot of cost, both on the spare part side but also on mast climbing, troubleshooting, tuning and time to market.
Most vendors of radio equipment provide base stations that are close to multi-standard today. The missing piece is the filters that are used to define the frequency of operation for the base station. These narrow filters are normally located close to the antennas, which means it is costly and time-consuming to swap them if a site is assigned more spectrum after a spectrum auction or an acquisition of another operator.
To obtain a fully flexible and multi-frequency base station, these narrow filters need to come with support for remote tuning. This can technically be done with small motors that are used to adjust the cutoff frequency of the filters. Many companies have looked at this, but the technology has not yet reached maturity as the combination of price, size and quality has not yet been met on the motors that should control the filters.
The market for tunable filters is massive. There are several millions of base stations deployed world-wide today and the density will increase with 5G. Each base station comes with many filters and each filter requires many motors to be fully tunable. The future for motor controlled multi-frequency base stations is bright. What the market is waiting for is just the right kind of motors.