Empowering the Next Generation City-Scale Smart Systems

Abstract

The onset of Internet of Things and the increasing level of integration between edge devices, commodity computers and low power sensors is empowering the next information technology revolution -- city scale extensible smart systems. Our team is working on such a system called Transit Hub. It puts accurate and real time information about optimal travel options into citizens' hands by processing information from transit system schedules, automated vehicle locators, emergency alerts, riders' smartphones, and dozens of other sources. Among other analysis, anyone with our free app1 can compare travel times using all available options -- biking, walking, public transit, driving and more. We are also working with the Nashville's Metropolitan Transport Authority (MTA) to provide them with a dashboard for monitoring and decision support to control the transit schedules and maintenance.Extensibility and smartness are two pillars of this emerging concept. Extensibility implies the flexibility to grow or shrink both in the set of services supported by a group of smart applications as well as the set of harware resources being used to provide those services. This flexibility is critical to create a feasible architecture that can be self-reliant over a long time. Smartness implies the ability to provide services that become better over time as more data become available. Data collection, dissemination and rapid processing and analysis of that data are central points of this vision. Elastic and multi-tenant computing resources have existed for a while. However, interactions with noncomputing resources and sensors are rarely an issue addressed by cloud computing where everything is virtualized without consideration for the management of resources that are not part of the computation platform.Constructing reliable applications across these dynamic smart systems is challenging. First of all, the set of available physical resources that are part of the platform can change over time either due to physical constraints or due to failures or due to addition or removal of resources. Second, by design the system is heterogeneous wherein different sensors and computation platform provide dynamic data and different resources and capabilities. Third, security cannot be an afterthought. All information, especially those related to critical data, needs to be controlled under an overarching security policy. Fourth, resilience is essential. Anything can go wrong at any time. The platform must be able to handle both internal faults as well as dynamicity of the environmental while ensuring that the application properties and requirements including strict timing requirements are met.In this paper, we are going to describe the Transit Hub system that addresses the above mentioned challenges. The details will include Transit Hub's different sensors that provide dynamic data, the different data analytics we perform on the learned transportation model, the decision support system that provides recommendations to the city authorities to control the transit system and also the core computation architecture behind the Transit Hub. It is built upon a novel architecture that provides strong isolation capabilities between different applications, supports dynamic reconfiguration including redeployment of resources of application components in response to changing environment conditions and component failures. Lastly, we will evaluate the architecture that has been optimized to make efficient use of available resource while executing long running analyses.