Poster Abstract: Reducing Tail Response Time of Vehicular Applications

Today's vehicles are sensor-rich but computation-poor. To better assist drivers, current vehicles have a large number of diverse sensors, for instance, the 2017 Ford GT has over 50 built-in cameras and sensors that can determine speed, location, humidity, occupancy, mechanical positioning, and a wealth of other data. However, modern vehicles have little general-purpose computing capacity due to cost, maintenance, and survivability concerns. For instance, vehicle manufacturers aim for vehicles to last for 20 years, and any general-purpose computing platforms would become obsolete or need maintenance many times during that lifespan.