SimCity: An Experimental Platform For Processing Context Inconsistencies And Its Demonstration

We are in the era of pervasive computing,or ubiquitous computing.Massive smart phones,tablet PCs and other computing devices equipped with various sensors can sam-ple information about the surrounding environment.One kind of applications called context-aware applications are capable of utilizing contextual information to perceive environmental changes,and adapt their behavior to these changes to provide targeted services for users.However,taking into consideration the complexity of the envi-ronment and the limitation of sensor accuracy,the collected contexts usually contain noises.These inaccurate noise data will mislead the application about the accurate per-ception of the environment,and then cause the application to run abnormally.To ensure that the application is using the correct contexts,we can add a mid-dleware between the application and the sensor aquisition module to detect and resolve inaccurate contexts.Currently one promising approach is constraint checking.Accord-ing to the characteristics of the surrounding environment,we design a set of consistency constraints for the sampled contexts.Each constraint describes a property that checked contexts must satisfy.If the checked contexts violates a constaint,we believe these con-texts contain an inconsistency.The next step is to select a proper resolution strategy to resolve this detected inconsistency.Due to the importance of inconsistency processing,we designed and implemented an experimental platform named SimCity for processing context inconsistency and its demonstration.It combines hardware and software,and can effectively perform verification on consistency checking and resolution.The primary contributions of this article are listed as follows:1.We built a physical experimental platform,including city buildings,streets,Lego EV3 bricks,Lego EV3 infrared sensors and bluetooth device Zenwheels Micro Cars.This physical mini city utilizes lots of IR sensors to continuously generate real contexts,and provides a real and adequate data source for verifying context inconsistency processing.2.We implemented a software system with multiple interactive modules.It can(1)automatically maintain the connections with all devices and attempt to recon-nect the device after detecting its disconnection;(2)automatically use infrared sensors to generate contexts and submit these to the consistency checking and resolution;(3)constantly control cars by keeping generating tasks;(4)analyze traffic conditions and schedule cars to avoid crashing into eache other;(5)switch the system state depending on the sudden incident,and improve system robust-ness by performing Reset,Relocation and Suspension.3.We provided an easy-to-use user interface.Users can influence the behavior of SimCity by controlling cars and creating tasks.They can also enable different scenarios to observe the effects of different inconsistency processing.4.We evaluated the performance of SimCity through several experiments.The re-sults show that the success rate of resolving inconsistencies has reached 84%.For the unresolved inconsitencies,we use Relocation to handle these,and the success rate of Relocation has reached 87%.When verifying one kind of consis-tency checking which performs constraint checking upon context change,SimC-ity found a new situation in which the detected inconsistency is a false positive,compared to two verification methods,using mock contexts and recording real contexts.This shows that SimCity has a better ability to verify inconsistency processing.