In the current digital age, massive amounts of data are generated in many different ways and forms. The data may be collected from everything from personal web logs to purposefully placed sensors. Today, companies and researchers use this data for everything from targeted personalized ads based on social data to solving important scientific problems that may help future generations of word citizens. Regardless if measured in monetary profit or other measures, the value of this data has proven valuable for many purposes and has led us into the Big Data era. Due to the large volume of data, Big Data requires significant storage, processing, and bandwidth resources. To date, the Cloud provides the largest collection of disk storage, CPU power, and network bandwidth, which makes it a natural choice for housing the Big Data.
This paper presents a resource-efficient protocolfor opportunistic routing in delay-tolerant networks (DTN).First, our approach exploits the context of mobile nodes(speed, direction of movement and radio range) to estimatethe size of a contact window. This knowledge is exploitedto make better forwarding decisions and to minimize theprobability of partially transmitted messages. Optimizingthe use of bandwidth during overloads helps reduce energyconsumption since partially transmitted messages are uselessand waste transmission power. Second, we use a differen-tiation mechanism based on message utility. This allowsallocating more resources for high utility messages. Moreprecisely, messages are replicated in the order of highestutility first, and removed from the buffers in the reverseorder. To illustrate the benefit of such a scheme, global accu-mulated utility is used as a system-wide performance metric.Third, we present a combined fragmentation/redundancyscheme which not only improves delivery ratio but also, ifinfrastructure is available, allows messages to be completedby pulling dropped fragments.Simulations illustrate the benefit of our model and showthat our scheme provides lower overhead and higher deliv-ery ratio, as well as higher accumulated utility comparedto a number of well-known algorithms including Maxpropand SprayAndWait.