Before reading the description of the bare idea presented below, written in 2009, it is worth taking a look at the article in Membrane, which… refers to an already existing one project (project, but not its implementation) of this kind. (Update) — it’s been working analytical system And also ideas and developments related to the investigation of the riots in London. In other words, there is practically an evolution from disjointed surveillance cameras into a unified tracking system. system Welcome to the new world.
So: An autonomous system powered by solar energy or another renewable source, equipped with sensors ranging from high-resolution optical stereo cameras, microphones, ultrasonic and microwave radars to RFID analyzers (the set of sensors may vary), large memory, a pattern recognition system (from biometrics for living individuals to the identification of inanimate objects), and communication with neighboring similar systems organized on a P2P basis. The system must understand geography: its own coordinates and those of the area it observes, as well as the points of intersection and “overlap” of the operational zones of neighboring cameras.
The system operates on the following principle: upon detecting an object within its field of view, it sends data from its own sensors to neighboring cells, starting with the one that is most likely to have been vacated by the object, with a request: “Have you seen this?” The response will be the reaction of one or more cells saying, “Yes, we saw it, it passed by at such-and-such a time, assigned identifier such-and-such.” The system links the existing image of the object with the received identifier and tracks its movements (recording them) until the next cell. After that, it transmits the data about the object to the next cell.
The “clock” distributes all the recorded information to the surrounding cells in pieces, so that it can later be restored in case of a failure of either the original cell or a certain number of cells that received part of the information from the data-hosting cell. Additionally, the “clock” utilizes the computational resources of the entire network for its operation. Essentially, it’s P2P. It is impossible to forge or delete information, and its distribution cannot be stopped.
If nothing is known about the object to the neighboring cells (the object has appeared for the first time in this cell), then this cell, after consulting its neighbors, assigns an identifier to the object and spreads the profile of the new object through “word of mouth.” If any of the sentinels recognizes it, the duplication of data is eliminated, and the object begins to exist with a single identifier. As long as no other sentinel has recognized the object, a “case” is opened for the object, and its movements and activities are recorded in the memory of the sentinel, shared with neighboring cells, and all information is distributed across the network (that is, neighboring sentinels act as peers in a P2P network) — in general, everything is as described above.
The system user can thus access complete information about all tracked objects through the interface of any “watcher” located in the cellular network. This includes their history and evolution. Moreover, the system will be extremely reliable—seriously damaging the data would require destroying half of all the “watchers” at once, and making unauthorized changes to the data will also be impossible due to its multiple duplication and logical connection with the data of neighboring “cells.”
A set of such “sentinels” effortlessly and reliably addresses any issues related to monitoring and securing an area, without the need for any “nodal” points or “dispatch centers.” The “sentinels” will sort everything out right after they are turned on. By the way, video surveillance systems not only transmit images to the operator but also record them on their internal memory in a “cyclic overwrite” mode. So, it’s just a step from a single recording camera to the idea of a coordinated swarm.
The standardization of the “watch” and the information exchange protocol will enable all its functions to be implemented “on a single chip,” making it quite affordable. Such watches could be purchased by private individuals for their own security, and their effectiveness will increase as more neighbors install similar watches, which will automatically find and identify their automatic “colleagues.”
The amount of memory required for a watch to operate for an extended period is already achievable in a relatively compact form.