Lowering the Barriers to Large-Scale Mobile Crowdsensing

Motivation:

This paper discusses three major obstacles of achieving large-scale mobile crowdsensing, and then proposes a deployment model to leverage the existing programming frameworks for crowdsensing applications.

Main points:

-- Crowdsensing applications have three major barriers, including heterogeneity of sensing hardware and mobile platform, the burden for users to install a application, and the growing of network bandwidth demands.

-- This paper proposes a deployment model. It is a 3-tier system architecture that separates data collection and sharing, removes app installation, and decentralizes processing and data aggregation.

-- For the deployment model, there are still some challenges, including the complexity of process migration, lower performance of inter-VM communication, the need of advanced network reconfiguration, and a lack of consensus on standard sensor data descriptions.

-- Deployment model depends on two assumptions. 

• First, it relies on distributed cloud infrastructure near the user.
• Second, it assumes that there is a standard API between proxy VM and application VMs to transfer data.

Trade-offs:

-- The deployment model depends on distributed cloud infrastructure nearby. But it can not be guaranteed in all the situations, which adds a restriction for this model.

--  The paper makes an assumption that a standard API exists between proxy VM and application VMs. But it doesn’t provide detailed description about the feasibility of this method, like how the programming frameworks mentioned in the paper can be implemented in this deployment model.