Pilot: Technical Post-Mortem

As the rest of the team returns from Haiti, our technical team (read: me) is taking a step back and understanding what we learned from the adventure. I did not get the chance to go out to Respiré, however I got the chance to be there virtually through the technology we were testing at the school. Despite a number of setbacks, I consider the pilot program to have been a great success from a technical perspective.

Rebecca and Isabel walk amongst the class, showing the students the ebooks we loaded onto the e-readers.

Rebecca and Isabel walk amongst the class, showing the students the ebooks we loaded onto the e-readers.

Overview

The technical goals for the pilot at Respiré were open-ended, as we wanted to test as much as we could to ensure we could supply ebooks to classrooms in a sustainable way. In the end, as we had a good number of Android devices, donated by Datawind to the school, we wanted to provide sufficient software and logistics to deal with the irregular internet and power at the school. We investigated a number of different reader applications for Android. A key goal for us at the time was that we could support DRM on ebooks as it was a strong requirement for a number of publishers. Not all content required DRM: Open Educations Resources (OERs) and some books from publishers of Haitian Creole content were provided as un-encrypted EPUB and PDF files.

We considered building our own solution from scratch, and weighed a number of options to make this happen. In particular, building a custom e-reader meant supporting DRM for commercial educational resources. In the end we had a great opportunity to partner with an existing app platform that supported DRM – Page Foundry offered not just their Android application, but also their library backend.

Me and Maria Rabinovich, UI Designer, planning the next phase of the Library For All app development

Me and Maria Rabinovich, UI Designer, planning the next phase of the Library For All app development

The Hub

The key challenge was that the school has extremely low bandwidth internet – the only access is provided through 3G USB sticks. To adapt to this problem, the intention was to also deploy an intermediate “hub” that would act as both a WiFi hotspot for the school and as a book cache that could reduce the need to transfer books over the internet. Rather than have each device download a book from the internet each time, the hub acts as an intermediary that stores each book on its first download, and then provides the book locally on each subsequent download to a device.

We wanted something that was self-contained, reducing the chance of pieces going missing or failing. We settled on an Intel NUC PC, which can act as a WiFi access point and a full-powered PC with a hard drive. This results in only 2 needed parts: the NUC hub and a USB 3G internet stick.

In order to ensure that the hub could withstand regular power outages, we needed to ensure that the entire network could quickly re-configure itself the moment the power turned on. In addition, we had heard reports that the internet devices could stop working occasionally. Before sending the router out to the school, we added software that would repeatedly check to make sure all of the various pieces would be up and running so long as there was power.

One particularly interesting piece was how to manage the hub. We couldn’t expect school administrators to do technical work so we set up a way for our technical team to log into the hub remotely to add functionality. This was an interesting challenge, partly due to the limited internet but also as we didn’t have the USB 3G stick to test everything out with. Amazingly, connecting to the hub from the US

Tanyella made sure the tablets were loaded with the content we wanted to test with the students.

Tanyella made sure the tablets were loaded with the content we wanted to test with the students.

The Conclusion

The biggest setback for testing the technical requirements for the pilot was that Page Foundry’s app required a bit more to get working in the field. The app required internet in critical actions, and we couldn’t update the app in the short time we were in Haiti. To test classroom scenarios, we used our backup solution: a simple book reader app and books manually copied to the device.

Despite this, we learned so much from the trip that we’re still working through all of it. We learned that we need to adapt to how different telecom providers supply internet, to ensure that we don’t run into delays connecting the hubs. We learned that remote management of hubs like the one we tested at Respiré needs its own infrastructure.

But critically, we learned that solutions using electronics require better infrastructure – electricity in particular. The tablets could be used for several hours, and while there was some electricity available for recharging them, it wasn’t sufficient to provide books to schools like Respiré. Extended solutions, such as tablets that can last several days without a charge or solar charging stations will be critical in ensuring the tablets remain in working order, and students have constant access to great resources, all the time.