Homework for February 14

http://www.ndsl.kaist.edu/~kyoungsoo/papers/dtp.pdf

My paper of choice was "A Disruption-tolerant Transmission Protocol for Practical
Mobile Data Offloading" by Younghwan Go, YoungGyoun Moon, Giyoung Nam, and KyoungSoo Park. This article provides background and practical application solutions, where a device can benefit from multiple and overlapping available data transfer technology, thus benefiting the user with faster response while benefiting the network with reducing the load on a much costly wireless mobile data transfer medium.

Because our project is going to involve mobility, we want to pay particular attention, how our solution will tackle network connectivity issues. According to the authors, most current applications has to solve disturbance issues in the code itself, instead of relying on the provided framework. The essence of their approach is condense down as following: 

"Instead of binding the connection on the four connection tuples (source and destination IP addresses and port numbers), DTP binds the connection to a flow ID that is agreed at the initial connection setup time and does not change during the connection lifetime. When a mobile host moves to another network, it can resume the connection with a new IP address and a port number by cryptographically attesting that it owns the flow ID of the connection."

One of their key concept is, that such Disruption Tolerant Protocol shall be transparent for both the application and the network. While the developers main goal was to make a data connection "seamless", one of the problems is that this protocol is not optimized for the fastest response time; so in this setup we just want to be aware of their idea, not necessarily implement it in our project.

In the article, the developers were concerned about large-file transfer (100MB+), but since we are not planning on having such files / communication, we do not have to worry about resume options.

From the energy consumption standpoint, DTP actually uses less energy on the mobile device in every scenario, if we compare it with TCP (even if there is no disruption in the connection).

This transport protocol is well worth looking into, as our project shall address mobility issues. If there is a possibility for using 3G/4G and Wi-Fi combined, the better our service could be. The authors developed an Android SDK for the protocol, and according to their claims, current applications can be ported with minimal effort. (As of today, I was unable to find the API for download, however it is highly possible that it is published somewhere)

Thanks for reading
--
Balazs Makany