A research conducted by American University Carnegie Mellon in Pittsburgh, says the nervous system of the fly could help improve the design of wireless networks. Researchers have been inspired by the way in which the fruit fly organizes its tiny hair like structures to feel and hear the world to improve the design of distributed computing applications.
The cells of the fly’s nervous system is organized so that a small number of them function as leaders to provide direct connections with different nerve cells. Researchers have developed the same kind of framework for distributed computing networks that develop everyday tasks such as Web search or control of an aircraft in flight. But the method that evolution has given the fly’s nervous system to organize is much simpler and more powerful than any created by humans.
The researchers used information on fruit flies to design a computer algorithm that is distributed and discovered qualities that make it particularly adaptable to networks where the number and position of the nodes is not fully established. These networks include wireless sensors, such as environmental monitoring, where sensors are scattered over a lake or waterway, or control systems for groups of robots.
In the computing world, a step towards the creation of distributed systems is to find a small group of processors that can be used to quickly communicate with other processors of the network, what theorists call a maximum independent set (CIM). Each processor in the network is a leading member of CIM, or connected to it, but leaders are not interconnected.
A similar organization occurs in the fruit fly, which uses tiny whiskers to detect the outside world. Each whisker is developed from a nerve cell called the sensory organ precursor (POS), which connects to nearby nerve cells, but not with other POS.
For three decades, scientists have wondered how a network processor can choose the members of the CIM. During the larval and pupal stages of fly development, the nervous system uses a probabilistic approach to select the cells that become POS.
In the fly, however, cells do not have information on how they are interconnected. As more cells are self-selected as POS, send chemical signals to nearby cells that inhibit these cells also become POS. This process continues until all cells are either POS or next to a POS and the fly emerges from the chrysalis.
According to investigators, on the fly the probability that any cell self-select increases not as a function of connections, as in the typical CIM algorithm for computer networks, but as a function of time. The method does not require advanced knowledge about how cells are organized. Communication between cells is as simple as can be.
The scientists created a computer algorithm based on the system of the fly and tried to provide a quick solution to the problem of CIM. In this sense, the authors note that the operation time was slightly higher than current methods but that the biological method is efficient and robust because it requires many assumptions, which makes the solution applicable in many more applications.