Garbage Collection for Edge Computing

Abstract

The shift towards edge computing is bringing data processing and storage closer to the edge of the network. This makes it desirable to use productive modern programming languages, like Python and C#, to program edge devices. Modern programming languages mitigate the added complexity of edge computing by abstracting software developers away from tedious tasks like freeing unused memory. But these languages rely on garbage collectors that impose high overheads and introduce unpredictable pauses, so they are rarely used in small embedded systems that make up the majority of edge devices. We propose a novel hardware garbage collector that addresses these problems to unlock the benefits of modern languages in edge devices.