Flocks, herds and schools - the roles of individuals in distributed sensor systems

对工程师来说，高度分散的“蜂群”系统的吸引力在于，整体超过了各个部分的总和，也就是说，蜂群的集体行为比每个蜂群成员的行为更复杂、更有用。研究表明，个体遵循的简单规则导致了自然界中复杂的“社会”，例如蚂蚁和白蚁。除了空间分布之外，这种方法还提供了健壮性、冗余性，特别是并行信息处理。在工程方面，电子和光电子设备的大规模生产使由成百上千个单独元件组成的系统在技术上和经济上都是可行的。该项目建立在约克大学视觉传感器研究(Visor)实验室在分布式系统方面的丰富经验基础上，特别是图像信息的获取和综合处理。该项目的目标是开发和实施片上、传感器级别和系统级别的技术，以便根据场景和其他动态变化的环境影响来动态优化系统资源分配。这里的资源涉及传感器选择、分辨率、速度、数据简化和分析，以及通信方法、带宽等。为此，我们将利用自然系统中的概念，如鸟群、鱼群或白蚁群落。特别是，我们需要了解如何为单个传感器制定简单的行为规则，这些规则导致系统的期望集体行为，这些行为在工程应用中是实用的。这项新提出的研究将开发对超低功耗成像系统有价值的技术-片上、生物医学应用、纳米和微微卫星，以及便携式消费产品。

Highly distributed 'swarm' systems have the attraction to engineers that the whole exceeds the sum of the parts, that is, that the collective behaviour of the swarm is more complex and useful than that of each member of the swarm.  Studies have shown that simple rules followed by the individual lead to the complex 'societies' demonstrated in the natural world by, for example, ants and termites. In addition to spatial distribution, such approaches provide robustness, redundancy, and in particular parallel information processing. In the engineering context, mass production of electronic and opto-electronic devices has made systems comprising hundreds or thousands of individual elements both technically and economically feasible. This project builds on the extensive experience of York University's Vision Sensor Research (VISOR) Laboratory in distributed systems, in particular the acquisition and integrated processing of image information. Examples include space-based meta-arrays of sensors and the current work on electronic compound eyes and other distributed sensor systems.The goal of the proposed project is to develop and implement on-chip, sensor-level and system-level techniques for dynamic optimisation of system resource allocation based on the scene and other dynamically changing environmental influences. Resources here refer to sensor selection, resolution, speed, data reduction and analysis, as well as communication methods, bandwidth, etc. To do this we will exploit concepts from natural systems, such as flocks of birds, schools of fish, or termite colonies. In particular, we need to understand how to develop simple behavioural rules for individual sensors which lead to a desired collective behaviour of the system that are practical for engineering applications. The  proposed research will develop technology valuable for ultra-low-power imaging systems-on-chip, biomedical applications, nano- and pico-satellites, and portable consumer products.