The Operational Challenges of mmWave Satellite Networks

Founded in 2020, Atheras Analytics has developed a suite of software tools that enable satellite operators to optimize the design and operation of their high throughput satellite (HTS) services in Ka and Q/V-band. I spoke to John Yates, Managing Director at Atheras, to find out more about the design considerations that must be taken into account at these frequencies and how their AI based, SaaS-delivered tools assist with this.

Q: Can you describe the challenges of operating in the mmWave frequency bands?

A: Most people in the industry are familiar with the challenges of using higher frequencies such as Ka-Band & Q/V-Band. They are widely used as they can deliver the high data rates that people need and can meet today’s hunger for demanding applications. The downside is that these frequencies are much more sensitive to weather attenuation than lower frequencies such as C & Ku-Band.

Traditionally, we have dealt with rain fade by using fade mitigation techniques such as ACM (adaptive coding & modulation), uplink power control or fade margin. However, with Ka- and Q/V band we can’t do that as the attenuation is so great. To overcome this, diversity gateways are used, where you have a primary gateway and a secondary gateway. If one gateway becomes affected by weather, the operator can switch over almost seamlessly to the backup gateway, and maintain the service. Whilst it is a solution to the problem, it does mean that it doubles the gateway cost, but in the grand scheme of things, it is a cost that operators are prepared to absorb.

High throughput satellites can carry from 200Gbps to 1Tbps (bits per second), and therefore an operator will require gateways in the tens or even hundreds. In providing a backup gateway for each, ground segment costs double immediately. This results in enormous cost.

The best alternative to using 1:1 redundancy is Smart Gateway Diversity. Using this approach, an operator could have, for example, fifty primary gateways and 10 or 15 backup gateways as it is highly unlikely that all of them will be out of service simultaneously due to weather conditions. The problem is that the operator doesn’t know which gateway they will switch to, so they cannot be synchronized. The customer databases have to be synchronized before the switch and this can take up to 15 minutes. By adopting smart gateway diversity, this can save 30-40% of you CAPEX yet if you don’t have the means to predict these outages, every time a switch is carried out, customers are going to lose their service.

Atheras Analytics has developed a machine learning algorithm that tackles these challenges. Utilising high resolution enhanced weather forecasts from third parties, we apply a specially designed algorithm which predicts, with a high degree of accuracy, when those outages are going to occur up to 6 hours in advance. Then, thirty minutes before the outage, we send a message to the satellite or gateway operator and advise them that they are going to lose the traffic due to weather. Simultaneously, we are monitoring their entire network, so we can see at which gateway there is availability for the coming hours and suggest transferring the traffic there.

This advance warning gives the operator thirty minutes in which to prepare for the switch, synchronize the databases, all whilst the traffic is active, and then once this is done, they can switch to the backup gateway before the outage actually occurs, so they keep the availability up where their customers need it.

When an operator switches from one gateway to another, 95% of that time is preparation time, and then the actual switch itself takes less than 5% of the time.

Q: You have just described your operational tool, but you also have a design tool?

A: Correct. With the design tool, if you are planning a network, you may have a dozen possible sites and want to know which are the best locations for your gateways.

We take 5 years of weather data for the potential sites. We then analyze the availability of these sites so that they can then be ranked. It is important that we only look at the last 5 years of data as some models take statistical averages from the last twenty to forty years and these don’t take into account climate change.

We apply the data to our algorithm and create a “league table” from this information showing the best to the worst availability. We then carry out a correlation between the gateways, as there could be two gateways that have good availability but if they are subject to the same weather outages at the same time they should not be selected. We are looking at gateways that have a high degree of de-correlation, so they are not both affected by the same weather system.

Q: So, you may not want to have two in the UK, you might want one in the UK and one in Italy?

A: Potentially, but every region of the world has its own weather characteristics.  For example, in Northern Europe, you will find that all the weather fronts generally come in from the Atlantic, and are generally aligned north-south. So, it may be best not to place diversity gateways on a north-south axis, but east-west instead. For example, in the UK one gateway in West Wales and one in East Anglia might give you the diversity that need.  It’s different all over the world. This is just one example.

We can use our AI tools to determine the gateway locations with the best availability but the question is much wider than that and is not just about availability. There are also other considerations such as economic, political and regulatory. We might have come up with the best location from an availability point of view, but it might not be suitable due to other issue. If you want to serve a certain country, they might insist that you have a gateway in that country. Questions around locating a gateway in a politically unstable area also arise. Some gateways are also cheaper to operate than others. For example, if a satellite operator owns one site and rents another one, then it is probably more cost effective to put traffic through the site they own. We can feed all this data into our AI model and determine the most effective gateway locations for specific overall network availability taking into account all of these factors.

Q: It is not just taking into account the location but you can also take into account system design?

A: Absolutely, it is very much a collaborative effort working with the customer because there are many non-technical factors to take into account. A government or defense satellite operator would clearly want gateways on sovereign territory. It also depends on satellite footprint. This is one of the first things that you would look at and will determine initially where the gateways could go.

Q: Who is this tool targeted towards?

A: It is primarily targeting satellite operators that are operating in the high throughput Ka-Band and Q/V Band satellites. But even for a gateway operator, we can still give them advance notice of weather events at their earth stations. They could use this to take steps to remedy the issue or to send messages out to the customers to warn them that they will lose service. It is primarily targeted at satellite operators but is in no way limited to that.

Q: Do you see climate change causing more design considerations in the future?

A: We certainly see the effects of climate change in long-term statistical models where averages are collected over a long period of time. However, our 5-year approach reflects the real world much more. And as we see more impacts of climate change, our algorithm will just deal with this because we get the new weather data and the models automatically take it into account.

Q: You also have a third product called UTOPiA?

A: This product uses the same core algorithms as the other products but is very much focused on end users. There are tens of millions of satellite-based internet end users, and the majority of these will be consumers. If they suffer an outage due to weather, they will probably just make a coffee and wait for it to come back.

However, if you are a high value end user, with a mission critical application, such as financial institutions, mineral extraction operation, government, defense, utilities etc., then you need to know that you are going to lose your connection, as it will have a major impact on your operations. UTOPiA (User Terminal Outage Prediction Algorithm) ensures that users receive a notification thirty to forty minutes in advance via an email, text message or app to let them know that they are going to lose their service. This gives them the opportunity to take mitigating actions to minimize the impact of that weather outage. It could be backing up a database or switching traffic to a different circuit if it is available – this is down to the end user.

For users with mission critical application, where an outage really matters, this service is invaluable.

We intend to sell UTOPiA through satellite service providers, so that they can bundle it as a value-added service to their service offering.

Q: Are there plans for your tools to cover other frequencies?

A: Rain fade is less of an issue at the lower frequencies such as Ku-Band as you can normally manage it through traditional fade mitigation techniques, such as by building in some power margin, and using uplink power control and ACM. So it is surprising that we are seeing some interest from X-Band users, so we may train one of our algorithms at this frequency in the future.

We feel that this is because X-Band is predominantly military and for some applications there is pressure to reduce the size of the terminal. If you are reducing the size of the terminal, you are reducing your link budget margin. We can imagine a scenario where people are operating on the limit of the units’ capabilities because there is pressure to reduce the terminal size.

Q: What future developments do you have planned?

A: In 2024, we want to develop an algorithm for optical communications. Looking beyond RF, there is increasing interest in using optical laser communications for feeder links. This is much more sensitive and it doesn’t just have to address rainfall because even cloud can attenuate an optical link. We are doing some preliminary work on this and talking with some potential partners for us to develop this algorithm.

People have traditionally put optical stations in the Canary Islands on top of mountains, so you are above the clouds and you have clear visibility. But if you are using an optical station for communications and you are working with a 1 Tbps high-throughput satellite, then you also need to have a suitable internet connection to that gateway. This capacity is often just not available at the top of a mountain. It’s a matter of striking the balance between locating an optical station near the internet backbone to give ultra-high speed access but at the same time in an area that is not disadvantaged with excessive cloud cover. This would be one of the data points that our algorithm would look at.

Another reason why we are looking at developing a product for optical communications is the potential use of Quantum Key technology being the future of encryption. Quantum keys can only be distributed on an optical media due to them being linked to energy levels. You can distribute the keys on a fiber optic cable or a laser link, but you cannot put them on a coaxial cable. A fiber optic cable would be able to transmit a key for a maximum range of around 500-600 kilometers, after which it will lose its integrity. If you want truly global quantum key distribution, then you have to use satellite-based laser links, so if the world decides that this is the way to go for encryption, then satellites will be critical to that for that global distribution.