In an emergency and rescue (ER) scenario, e.g. in the aftermath of an earthquake, the regular communication infrastructure might not be available.Modern mobile devices can still, however, form a network between themselves. Such a network is commonly referred to as a Mobile Ad-hoc Network (MANET) and can be utilized for communication purposes. The nodes in a MANET are typically handheld devices with limited power and bandwidth.Due to the mobility and limited communication range of the devices, links between them may go up and down, A network with these properties is called a disruptive MANET.In a disruptive MANET there is no guarantee that we have a complete route from source to destination up at any one time.We therefore utilize delay tolerant networking techniques, such as a store-carry-forward scheme to ensure eventual delivery of packets.
We picture a scenario with two main locations and associated network partitions, namely the incident site and the off site Command and Control Center (CCC).Rescue workers at the incident site use head-mounted cameras to capture video containing information that needs to be transmitted to the personnel at the CCC.Transmission, being one of the main sources of power consumption can put a large toll on the network and lead to the depletion of the batteries of key devices at the emergency site.
Our goal is to reduce resource usage when conveying information based on video from an incident site to the CCC in a disruptive MANET. By extracting the information we are interested in from the video and discarding irrelevant information we reduce the power consumption and required bandwidth. Additionaly, we allow the most important information to reach the destination more quickly.We analyze the application scenario and come up with different approaches for doing this, one of them being to select key frames to represent whole video segments and transmit these frames as opposed to the whole video. The personnel at the receiving end can then decide which segments to receive first, based on what they see in the key frames. Another approach is transferring panoramic images, which are the results of stitching multiple overlapping images together to help give the receiver an overview of the site where the footage was taken. We propose a possible application for the workers at the CCC utilizing these and other approaches to give the personnel a best possible overview of the situation at the incident site.
We circle out one of the components of the CCC application, the image stitching component, and conduct an experiment to investigate the resource usage and power consumption of different approaches to utilizing image stitching. The results of this experiment shows that image stitching can help reduce the power consumption, amount of data transmitted and time spent compared to the transmission of video.