This is a dream about the possible future of mesh networking in mobile devices. If you prefer apple products, replace "Android" with "Apple", it really isn't too important.
A mesh network is a network where each device in the network talks to other devices nearby, which in turn relay messages until they reach their final destination.
There are three reasons why I wish that all mobile devices had a built-in, low-bandwidth mesh network. The first is that it would provide a free way to communicate for groups that don't want to pay a monthly cell phone bill. For example, consider preteens. They might have an ipod touch or a parent's old cell phone, but can't use it unless there is wifi. (Preteens love to text, which uses very little bandwidth.)
The second reason is during an emergency. Along with the mesh network, which would work as long as enough nodes still had power, special apps could be written to coordinate disaster response even without a reliable Internet connection.
The third reason is that I have a lot of devices that I wouldn't mind having an Internet connection on, but not enough to pay for them. For example, my laptop and tablet.
I'll now try to address the problems I forsee with a mesh network, and propose a way to overcome them.
In order to be useful, there need to be lots of awake nodes. To strengthen the network, each phone would come with a base station, which would be permanently plugged in the wall. This would act as a node in the mesh network as well as being able to connect to the wifi in that home, providing a bridge between the Internet and the mesh network.
This reduces the total load on the mesh network, since a packet can "jump" via the Internet. To prevent abuse, a packet would always route through its source base station. So for example, if person A is at the home of person B, and wants to look something up on wikipedia, a packet would go from person A's phone to person B's base station (or via the mesh network to another base station in a home nearby). It would then tunnel through the Internet to person A's base-station. That base station would then re-send the packet out to the Internet to wikipedia.
Said another way, the base stations would all be connected to a decentralized virtual private network (VPN). This makes the mesh network have connectivity from city to city.
While this seems obtuse, it is designed to stop someone from being able to generate traffic to websites from someone else's IP address. Some rate limiting would also be present at each base station to prevent abuse.
Also, because a base station is required to communicate with the Internet, users will be encouraged to plug them in and keep them working, which greatly strengthens the network.
The first problem with a mesh network of this type is that it needs to have enough range to be able to communicate over long distances, so that the mesh can function in an urban or suburban environment. 2.4GHz, as is used by wifi, doesn't travel far enough to be useful. Instead, you would use the FRS band, which is around 467 MHz, or another similar band. This would give enough range to be useful, especially when combined with the base stations.
The mesh network should also be able to participate in high-bandwidth exchanges when the range is small enough to use the existing 2.4GHz hardware, perhaps doing wifi-direct. This would allow for video conferencing using the mesh network.
In order to preserve battery life, mesh networking hardware typically synchronizes wakeup times. In busy areas where base-stations are present, a device could determine that it doesn't need to participate in the network as a relay, and can sleep for longer periods of time.
There are obviously lots of challenges to overcome with building a universal mesh network, but if a hardware manufacturer can overcome them, I imagine that all other manufacturers would be forced to add compatible technology, which would eventually lead to a global, decentralized network that anyone could use.