I had a blog about a month ago talking about Android and how it is making mobile grid computing reality. The major points that I talked about were improved networking and CPU performance of the modern mobile devices combined with unified programming model delivered by Android - jointly bringing mobile grid computing into realm of reality.

Since then I had several discussion about it and I have several interesting follow up observations.

First of all, the devices in mobile grid are profoundly different from those found in traditional grid computing environment. While rack mount blades are completely faceless and void of any human interaction - mobile devices like phones and PDAs are almost an extension of a human being - people interact with phones constantly, pressing button, snapping pictures, listening to music, making calls, watching movies, etc. What I find fascinating is that if you extend this idea of mobile device being the extension of human's capabilities you can think of mobile grid consisting of human nodes... Indeed, every person with the mobile device is a node in the grid and his or her device is just a connection to the grid.

In this sense, mobile grids open up completely different types of applications to be run on the grid. Some of the obvious choices are human-based image recognition, short and massively parallel tasks, new types of security (requiring simple human interaction from a group of people), etc.

Second of all, mobile grids will require new business models or approaches. I can clearly see, for example, large operators like AT&T or Verizon offering discounted or free services in exchange for having a mobile device participate in the grid - while reselling this processing capacity to other businesses. Let's say you get free 1000 any-minutes if you let your phone to crunch some small tasks while you are not talking... not a bad deal!

Overall, I predict that mobile grid computing will become a new trend in distributed processing in the coming years.