Traditional analytic systems take a centralized approach whereby all captured data flows to central analytic servers dedicated for analytic processing. Although such a solution offers a high amount of resource availability for processing, such a solution poses two major barriers in terms of scalability in a digital transformation.
High Bandwidth Requirements
The majority of unmanned system analytics are computer vision based (or other high bandwidth forms of data), which require high bandwidth data streams to be transmitted from each unmanned system in use to a central location for processing. This puts severe stress on the network infrastructure especially as the number of unmanned systems scales up.
High Processing Requirements
At low scale a dedicated analytics server offer high processing resource availability. As scale increases processing resources diminish especially in concurrent processing scenarios. Scaling up means investing heavily in expensive high resource servers.
The emergence of powerful, low-cost, and energy-efficient processors is quickly making it feasible to incorporate analytics capabilities into embedded system. The ability to process data at the networks edge (i.e. the unmanned systems themselves) provides a highly scalable distributed computing architecture for digita transformation and adds a level of intelligence to unmanned system rich for exploitation in terms of automation, autonomy, and swarm collaboration. Moreover, high bandwidth data transmission is no longer necessary for analytic processing, instead low bandwidth processed data (e.g. identified rust and corrosion snapshots) with respective metadata (e.g. gps, timestamp, and etc.) is transmitted for further processing and fusion. As unmanned systems are equipped with the necessary resources for embed processing no further investment is necessary when scaling up your digital blueprint.