Edge caching for cognitive applications

#The present invention relates generally to a method, system, and computer program product for caching within a communication network. More particularly, the present invention relates to a method, system, and computer program product for edge caching for cognitive applications in a communication network.

#In environments where network bandwidth is limited, a network cache can be used to improve the effective response time between a client and a server. A network cache is data storage that is located closer to the requester than a server that originally contains the requested data. Caching has been used effectively for applications such as web-applications, databases and computer processing. Web caching utilizes temporary storage for storing web documents, such as HTML pages and images, to reduce bandwidth usage, server load, and/or perceived lag.

#Although web-caching is used frequently, caching has also been used extensively in other distributed systems, such as information centric networks. Caching performs well in these environments because a request often includes a unique handle such as a Uniform Resource Identifier (URI). The handle can be used as a unique name that is associated with cached data. Caching works by implementing a proxy or gateway which allows requests from a client to pass through to a server. When the proxy receives a request containing a handle that has not been previously encountered, the proxy sends the requests to the server, sees the resulting response and stores the result locally. When a second request with the same handle is encountered by the proxy in the future, the locally stored data is sent by the proxy in a response to the client instead of sending the request to the server to get a response from the server. Various caching implementations are concerned with how to manage the locally cached data, as well as maintaining consistency between cache and server data.

#Certain novel features believed characteristic of the invention are set forth in the appended claims. The invention itself, however, as well as a preferred mode of use, further objectives and advantages thereof, will best be understood by reference to the following detailed description of the illustrative embodiments when read in conjunction with the accompanying drawings, wherein:

#FIG. 1 depicts a block diagram of a network of data processing systems in which illustrative embodiments may be implemented;

#FIG. 2 depicts a block diagram of a data processing system in which illustrative embodiments may be implemented;

#The illustrative embodiments described herein generally relate to edge caching for cognitive applications. As previously discussed, cognitive applications are computer processing applications which take a signal from a sensor device and apply machine learning techniques to a set of training data to produce an output. In accordance with one or more embodiments, a caching protocol is implemented between a proxy server having an associated local cache and a cognitive server, in which the cognitive server returns not only the result of applying an input signal on a cognitive model, but also implements a handle generation function or encoding function configured to generate a unique handle for an input signal. In one or more embodiments, the handle generator function or encoding function is computed on each input signal received from a sensor device to create a unique handle for the input signal, and the proxy server stores the result received from the cognitive server in association with the handle in the cache.

#In the embodiment, when the proxy server receives a new input signal from a sensor device, the proxy server computes a new handle for the new input signal using the same handle function used to compute the handle for the initial input signal, and determines whether the new handle matches the handle stored in the cache using predetermined matching criteria. If a match is found, the proxy server using the result associated with handle as the result for the new input signal. In one or more embodiments, the hit rate of the cache is improved through use of the unique handle. If a match is not found, the proxy server forwards the new input signal to the cognitive server, and the cognitive server computes a result and a handle for the new input signal. The cognitive server sends the new result and new handle to the proxy server. The proxy server then stores the new handle and new result or answer value within the cache associated with the proxy server. In accordance with one or more embodiments, a state of the cache is changed because the storing of the answer value modifies whatever is residing the cache prior to the storing. When a new input signal is received by the proxy server having a computed handle value matching a handle value stored in the cache, the proxy server can use the answer value associated with the matching handle value rather than requiring the cognitive server to determine the answer value for the new input signal.

#Accordingly, one or more embodiments described herein provide for an improvement in the effectiveness of caching for cognitive applications through the use of a proxy server to generate a handle for received input signals to assist in caching the input signals.