The Eyes Have It…Finally

by Gordy Slack

In this satellite map graphic of the Aravind network, green lines indicate links from the central hospital to rural vision centers in five rural towns. All distances are in kilometers. (Graphic by Sonesh Surana).

In many emerging economies, like India’s, advances in telemedicine can ensure that big-city healthcare is available even at the outskirts of town. But what about the millions of people who live beyond the outer reaches, where the “tele” in telemedicine does not yet reach?

UC Berkeley Professor Eric Brewer’s CITRIS-supported Technology and Infrastructure for Emerging Regions (TIER) project has found a way to bring broadband wireless to villages that, until now, have been off the tele-communication map. By modifying simple and readily available wi-fi technologies, the group has linked tiny local eye clinics in the southern India state of Tamil Nadu to bigger clinics, like the Aravind Eye Hospital at Theni.

In the past three and half years, the project has grown from just one clinic and one hospital to include now thirteen clinics linking up to three different hospitals. The clinics are providing videoconferences with eye doctors for about 3,000 rural patients a month, says Sonesh Surana, a computer science graduate student at UC Berkeley who has been involved with the project since its inception in late 2004.

Thanks to Wi-Fi connections enabled by the TIER group at CITRIS, patients in remote areas of India now have access to quality eye doctors through telemedicine.

If it were not for the remote clinics, very few of those patients would get any eye care at all. “Most of these people have never had a eye check-up before this” says Surana. “Having a doctor in a white coat on the other end of the videoconference actually looking them in the eye and paying attention to their ailments makes patients feel included and respected,” he says. And after such personal contact, the patients are much more likely to follow the advice the doctor gives.

In only about ten percent of the cases does that advice require a trip into town for surgery or some other procedure. Most of the other cases call for simpler prescriptions that can be filled by the local high-school-educated girls trained to run the village outposts, avoiding expensive and disruptive day-long trips into town for most of the clinics’ patients. Those who require cataract surgery or other procedures make appointments during the videoconference. They can also do on-line follow-up interviews after their surgery, saving more expensive and time consuming trips.

The TIER group used Wi-Fi cards based on the 802.11 networking standard, just like those found in most laptops. But the 802.11 standard typically limits that technology’s range to a “hotspot” reaching only about 200 feet, and the signal is broadcast out equally in all directions. By modifying the software and focusing the signal from routers with high-gain directional antennas, the group extended and narrowed its reach while retaining decent transmission speed. The networks they have set up can extend dozens of kilometers and operate at speeds of up to six or seven megabits per second.

The connection does require a direct line of site between stations, but obstacles can be bypassed by adding a relay on a tower, says Rabin Patra, another student of Brewer’s and a TIER member. Theoretically, adding relay stations could extend the connection to reach hundreds of kilometers, but the end-to-end latency grows with each relay and would eventually grow too long to make videoconferencing effective. The various nodes in the network can communicate with each other through the central hub, however, and this allows people in the participating villages to communicate with one another.

The system is reliable and fast enough for both videoconferencing and the transmission of clear ophthalmologic images and other data. It can extend into areas with no cell phone coverage, no cable, no wires, says Surana. All those things are key, but what is perhaps most extraordinary about this technology is its price.

The onetime Wi-Fi equipment cost amounts to only $800 per link; after setup the system is very inexpensive to operate. It uses only about seven watts of electricity, which can come from a variety or sources, including solar. Once capital equipment cost is amortized over the five years, the cost for this vital service comes to about only $200 per year.

The TIER researchers work with the Intel Research Berkeley lab team on the Aravind Eye Hospitals project through an open collaborative research agreement between Intel and UC Berkeley.

After three years, the system is working smoothly; local villagers are able to repair and maintain the networks on their own. But the TIER group is still working on improvements. Surana, for instance, is trying to make the system more reliable and robust.

A number of different problems can take the system down. But to someone working at the local clinic or at the hospital, they all look the same: no connection. Surana is developing built-in mechanisms to let users on both ends know when the system is limping, so steps can be taken beforehand to avert a fall. And he is also developing ways to let users know if, say, the system is down because of a simple power failure, or if the problem is something else—and susceptible to fixing.

One option, in areas where service is available, would be to employ cell phones as a backchannel; if the power goes off or if something else goes wrong with the connection, the cell phone could send a signal to the nodes and base station to let them know what was happening. “Cell phones are low bandwidth and high cost,” says Patra. “But, for very short and infrequent calls, they could be affordable and reliable sources of very valuable information.”

Patra is also working with another student, Sergiu Nedevschi, on further optimizing the system’s efficiency and speed. Right now, both ends of the link are sending and receiving data packets in one-to-one ratios. So that the signals don’t collide in the back-and-forth, one end transmits for a few milliseconds and then receives for a few milliseconds. These pulses are precisely timed and synchronized so the two ends of the connection are coordinated. But the direction of data flow isn’t always so balanced; at times one end does a lot more sending than receiving and at other times the ratio may be reversed. So, Patra is developing software that can sense the ratio of sent data to received data and shift the amount of sending versus receiving time allocated to fit those ratios. Their simulations predict that they improve efficiency by between 25 and 100 percent, depending on the traffic pattern and network topology.

The Aravind Eye Hospital aims to have the most affluent one-third of its patients subsidize care for the poorer two-thirds, so keeping costs low for subsidized patients is key to its success. The cost efficiency of the wireless system enables the hospital’s innovative subsidy structuring to work.

The network exposes rural patients to the clinic, on the one hand, and allows most of them to avoid traveling to town on the other. The program has other side effects, too. Training local young people to administer basic eye care has increased overall awareness of health issues in the villages, says Patra. And the networks themselves can be used to bring movies and other entertainment and information into the remote areas.

A new federal Indian program aims to set up 100,000 information kiosks in India’s rural villages. For those areas beyond the reach of traditional phone lines and other wireless service, TIER’s low cost, broad-band wi-fi may be the strongest, cheapest method of connecting those kiosks to the broader world. TIER’s wi-fi is being used in projects in Africa and the Philippines as well, and could easily be applied in remote areas in the US, too, says Patra. A government agency in California is monitoring forests for fires using long-distance Wi-Fi.

“The US is pretty well wired,” Patra says, “but sometimes we need a way to go that final ten or twenty miles.”

However, even if the system does not spread far beyond its use in the Aravind project, it has been well worth the effort, says Patra.

When asked how many other computer science graduate students can say they have made it possible for 200 more people to see each month, Patra laughs and says the credit goes to the hospitals, not the programmers. “We just filled in the missing link.”