Pervasive and mobile systems are constrained by energy and performance issues when they operate in isolation. Yet, several studies have shown that smart devices are frequently co-located in proximity to at least one other device throughout the day, suggesting that devices can potentially collaborate to reduce the effort of resource intensive tasks, e.g., sensing, offloading, networking, storage, etc. However, merging the resources of multiple devices to work together is a tough challenge as it requires a common understanding of the context of each device. In addition, since smart devices function in multiple roles, understanding what really constitutes context becomes difficult as different types of contexts need to be modeled depending on the type of task. The goal of this workshop is to explore the use and effect of contexts on multi-device settings, including collaborative and/or opportunistic sensing systems. The significance of this research area is corroborated by the ever increasing emergence of novel self-organizing computing infrastructures that operate in multi-device environments, e.g., edge and fog computing, and the increasing availability of openly available data sources that capture context from multiple devices.
Context information is critical in multi device setups, e.g., for the formation of collaboration groups, and for negotiating responsibilities between the devices. Any misunderstandings in context can thus be extremely counterproductive for the collaboration between devices. Consider, for example, a computational task, e.g., game puzzle, whose execution can be accelerated by dividing computations between two or more devices. To distribute the task, we first need to understand the processing capacity of each device to ensure devices with insufficient resources are not harnessed for this task as they would turn into bottleneck for execution. Similarly, for energy savings purposes, we need to understand the CPU level of each device and the energy consumption of processing to avoid using a device that it is overwhelmed with processing and that will spend more energy due to busy CPU rather than processing. What makes this task even more complex is that a single device rarely can capture sufficient information about its context and execution to be able to provide the required information for other devices. However, given the huge amount of devices and their similarities, and the fact that applications are instrumented to collect and store data into logs, it is possible to envision opportunities for capturing these aspects by aggregating measurements from multiple devices.
Our workshop is the first to address the opportunities and challenges arising from the combination of two emerging domains: crowdsensing based context-awareness and collaborative multi-device computing scenarios. Our unique focus makes PerCrowd an exciting forum to discuss key research challenges and familiarize with state-of-the-art research in these topics.