HTTP Request/Response Filters

The filter engines are responsible for modification of HTTP requests and responses in order to reduce data traffic over the bottleneck link. Typical examples of filtering include reducing the color depth of images, clipping images, transmitting parts of audio files only, transmitting the first few words of each paragraph in a document, reducing the HTTP request header, suppressing requests to documents from a particular server, etc. Every user has his/her own HTTP request filter and HTTP response filter.

Invocation of the appropriate filter is accomplished by means of a rules database. A rule in the database specifies the invocation of a filter based on its document type, available network quality of service, size of the document, etc. The goal of the rules database is to allow adaptive filtering based on dynamic network conditions and data types. While the set of rules in the rules database is currently small, the architecture allows for more sophisticated filtering.

The HTTP request filter also notifies the user of the documents that are pre-fetched via a well-known URL. That is, it intercepts a well-known URL and formulates an HTML page which represents the pre-fetched items in the cache. This allows the user to see which URLs have been pre-fetched and manually set or modify the list.