In an ideal ubiquitous computing environment, computing services would be embedded in physical objects all around us, which would all work together seamlessly. As computing could be embedded in all sorts of physical objects, a myriad of different interfaces could arise for different computing services with similar functionality. However, software in ubiquitous computing environments must still be able to access computing services in the environment even when the desired service has an unfamiliar interface, otherwise this will be an obstacle to the seamless operation of a ubiquitous computing environment.
Interface adapters can provide a solution to this problem by transforming interfaces as necessary. And chaining them together enables much more flexibility without incurring a prohibitive development cost in creating all of the required interface adapters for direct interface adaptation. Unfortunately, an interface adapter is likely to be imperfect, so interface adaptation would often incur adaptation loss. This is even more of an issue when interface adapters are chained. To properly consider loss in the construction of interface adapter chains, a mathematical framework is required to analyze such chains.
We develop a number of mathematical frameworks to analyze the loss in interface adapter chaining, each building upon another. For each mathematical framework, we define algebraic objects and operations that express the loss and how it is affected by each interface adapter. These in turn are used to develop algorithms and prove complexity results for problems relevant to lossy interface adapter chaining.