Differences

This shows you the differences between two versions of the page.

--- sensor_array [2014/10/13 21:09] – [Abstract] koko
+++ sensor_array [2014/10/20 21:38] (current) – [Abstract] koko
@@ Line 2: / Line 2: @@
 ===== Title =====
-A Robust Sensor System for a new Artifical Life Framework
+Robust Sensing in a Distributed Environment
 ===== Abstract =====
-In this paper, we describe a system in the movable feast machine framework of Artificial Life. The movable feast machine (MFM) is a recent development in the artificial life community, intended to describe a more intuitive framework for computation using a type of cellular automata. Using this framework requires a great deal of engineering, as many standard behaviors expected from a typical computer do not come built into the MFM. We present a design for a set of MFM elements which construct a phased sensor array to sense incoming data. The system as a whole gathers directionally oriented data transmissions and tells us where the data is coming from. This can enable users to do things like analyze incoming data or to establish communications pathways. We show that the design of the system captures the overall directionality patterns of incoming data and stores that information, for use by other parts of a larger system.
+In this paper, we present a design for a set of Movable Feast Machine (MFM) elements which construct a phased sensor array to sense incoming data. The system gathers spatially-oriented data transmissions and tells us from where the data is coming. This can enable users to analyze incoming data, or to establish communications pathways. The design includes parameters which regulate the density and directionality of the array. Here, we show that certain parameters allow the data to be sensed accurately without too much data loss. Specifically, we illustrate the trade-off between data loss and sensing capacity as a function of these parameters.
+===== Methods =====
+Any sensor system needs to have some sort of target object to sense. In this work, our target objects are called Light particles. These are a very simple pair of light atoms, driving themselves forward in a given direction. The Light particles are emitted at a steady rate by Emitter atoms. The locations of these Emitters are the main things we want to be able to sense with our system. As such, each experiment described here has a predetermined set of emitter locations, making it easier to compare the performance of different parameter values, as we do here.
+The Light particles are what we should sense, but they must interact with another element if we want to do this. In this setup, we use a number of Box atoms for this purpose. Box atoms construct a small N * M rectangular box, which will serve as the effective range for our sensors. The Box atoms in this wall each record the number of Light particles which strike them from the inside, while also disposing of these particles by deleting them. Emitters outside of the box are ignored in this work to narrow down the scope of our analyses.
+The Box alone serves as a very rough sensor, but it is incapable of distinguishing several different directions from each other, depending on how close to the Box's wall a given emitter is situated. See Figure XXX for an example of this phenomenon. The naked Box-as-sensor will serve as the control experiment against which the performance of other designs is measured.
+One other such design involves the use of one or more Antennae.
 ===== Conclusions =====
 We have illustrated the design for a phased array system, implemented in MFM elements. This fixed array allows us to detect incoming data and determine where it is coming from. We showed that the density of antennae and directions that they point in have a critical impact on the quality of sensing we get from the system. This system is intended to be a useful component in a larger MFM system, such as a router in a communication network or an artificial eye in an ALife organism. Such a system does not yet exist, but this system's design will hopefully spur further development in this area. Future plans involve the construction of such a general communication system for MFM systems, enabling multiple MFM implementations to interact. This proto-Internet would enable MFM developers to construct even larger systems on top of it, expanding upon the MFM's promise of massive scalability.