Telehealth over satellite broadband confronts the issue of latency
The CSIRO has released a whitepaper on the potential for telehealth provision to remote parts of Australia using satellite broadband, setting out some guidelines on what is achievable within the limitations of the technology.
Written by Sarah Dods and colleagues from CSIRO's Digital Productivity and Services Flagship and launched at the Global Telehealth conference in Sydney this week, the whitepaper investigates the differences between the type of broadband connection different parts of the country will receive once the National Broadband Network is rolled out.
Telehealth has long been heralded as a way to improve access to healthcare for people living in rural and remote regions, but there are some limitations to the type of broadband these regions will receive that will have an effect on the telehealth services healthcare organisations will be able to provide.
One of the major issues is latency, the time delay that satellite broadband in particular experiences that may rule out certain components of telehealth, including perhaps high-resolution video conferencing and remote operation.
Under the government's NBN plan, 93 per cent of premises will connect with broadband using optical fibre and four per cent will use wireless, the majority of them in regional areas. For the remaining three per cent of Australians living in remote and very remote regions, broadband will be provided by satellite.
However, it is this three per cent that has far lower access to healthcare – and far higher incidences of chronic disease and hospitalisations – that could most do with telehealth, Dr Dods told the conference.
She said the differences between satellite communication and other kinds of broadband need to be taken into account when considering the development of telehealth applications and the services they can deliver.
“There is an assumption that broadband just works and if you can develop the service then it can just run over the infrastructure that is there,” Dr Dods said. “If you are designing a service, what you should think about is determining whether it is appropriate for different kinds of broadband delivery, particularly when it comes to satellite.”
For fibre connections, the peak upload speed in the future is expected to be 400 megabits per second (Mb/s) and 1000 Mb/s download. For wireless and satellite, however, it will be 1Mb/s upload and 12Mb/s download.
An interim satellite service was launched last year that is capable of peak download speeds of up to six megabits per second, with a long-term satellite to be launched in 2015 that will be capable of 12Mb/s. Upload speeds will be 1Mb/s.
For telehealth services with two-way information sharing, it will be the lower upstream bandwidth that will set the limitations on the types of services that can be delivered, she said.
“Bandwidth will change depending on the upstream and downstream specifications, but there is also this thing called latency,” she said. “That is how long it is going to take to get to the other end.
“The amount of time it is going to take to get from a [house or medical centre in a remote area] to the NBN point of interconnect is going to be about a quarter of a second, best case. This is because it has to go up to the satellite and back.
“Once you get into the backhaul, it is optical fibre and travels at the speed of light and you have a pretty short transmission time through the middle section. But when you get to the final section, whether it is a doctor's surgery or a hospital, you are going to have some more delay at that end. Then the response has to return along the same path.
”That is going to affect what kind of experience you get for some services when you connect to the other end of the network. If the information flow is one way, such as streaming an educational video, it doesn’t matter. For two-way interaction, it makes a difference.”
If the connection is between satellite and fibre, the latency will be noticeable in conversation, but if both ends are using satellite, there will be “two hops” and the latency will double, she said.
Communications satellites are in geosynchronous orbit, orbiting the planet every 24 hours from 36,000km above earth. To send a signal at the speed of light to the satellite and back takes about a quarter of a second.
With two-hop satellite connections, the latency can be one second in the best case scenario. “What it means is that if [a remote patient] is connected over satellite, and he wants to talk to a rural health service, which is also connected over satellite, then he has to count it four times,” Dr Dods said.
“The interesting thing in this is the human element. Generally our reaction time is a quarter of a second. If things happen faster than a quarter of a second, we think it is real time. If it gets longer than 250 milliseconds, then we start noticing a delay and an unnaturalness when we are waiting for somebody to respond.
“So we are down to looking at at least a one-second reaction time for somebody at the other end to respond to what he has been saying. That can feel decidedly odd. Imagine trying to do a remote consultation between a city specialist and a remote clinic over that. This is why we hear people talking about latency and why it matters.”
Dr Dods and her team have developed a framework that outlines what different elements of telehealth provision are achievable using satellite, and how health services might be able to put those components together.
She urged healthcare organisations to look at what aspects of telehealth they can provide to remote areas now and in the future, despite the limitations of satellite. These include old-fashioned telephone-based services, the use of social media for sharing content, asynchronous information sharing or “store and forward” – the transfer of images or clinical data to be used at a later date – sensor monitoring and video streaming.
“What we have then done is look at how much bandwidth you are going to need to do the different services, and what the interaction time you will need for them. If you look at sensor monitoring, you only need minutes, but if you are going to look at remote operation, you will need it to be very fast.
“With low resolution video conferencing, that will have a significant delay but it will be fine if you are just giving some advice to a person and telling them what to do, but it may be a really bad thing to use if you are offering a telepsychiatry consultation and someone is recounting a fairly traumatic episode and you are trying to provide some real-time empathetic response.
“You need to think about these elements. If you are at a remote clinic and you want to provide home-based monitoring for chronic heart failure, you might have some simple sensors, some video streaming for education and some low-resolution video conferences – these are choices that you will need to look at to deliver the outcome you are looking for.
“If it is a city hospital providing the service, they have fibre so it is quite fast. If it is a rural clinic going to someone's home, you are going to have a two-satellite hop and about a second or more latency you need to design for in your system.”
Dr Dods said that while bandwidth will improve, latency probably will not. “The interaction timescales aren't going to change in the same way [as bandwidth]. I don't think we are going to be changing the speed of light anytime soon and I don't think we are going to be changing human reaction time.
“We are particularly interested in human interaction and how we can adapt to those limits, but there are some limitations that go against nature so we are going to walk away from them. Telesurgery over satellite is not going to happen in anybody's lifetime. You really have to think about that interaction aspect when you are looking at putting services together.”
Posted in Australian eHealth