Contextual Multi-Device Delivery

W3C Delivery Context Workshop Position Paper
Venu Vasudevan, Motorola Labs (
In the heterogenous post-PC device ecosystem, the web delivery infrastructure needs to be smart enough to contextually adapt content delivery to provide a satisfactory user experience. Current approaches to this problem include content transcoding architectures, and the maintenance of device-specific content versions. These approaches miniaturize the service to fit into a smaller (than enterprise class) device. A promising, complementary approach that is recently receiving greater attention  is multi-device delivery.

Multi-device delivery hinges on the observation that a mobile user's computational neighborhood is substantially more capable than any single mobile wireless device on his person. Instead of reducing content modalities to what is accomodated on a single device, multi-device delivery attempts to make use of all the modalities available in a computing neighborhood. Successful multi-device delivery sacrifices some service cohesion (from a user perspective) for greater richness in services, as the service is now delivered over combinations of devices and access networks. Variations to the theme range from applying  multi-device media delivery in the context of a single web retrieval [MyVV], to its applications in the context of multiple logically cohesive web retrievals [Fox], to work (from Ericsson) on coordinated delivery across multiple devices owned by a single user [Hers]. Ideas in this space should be applicable to multi-modal devices, as these devices can be viewed as a computing neighborhood packaged into a single device. Other standardization proposals in the Wireless World Research Forum [Raat] also point  to adaptive multi-device delivery as a priority.

The architectural framework for multi-modal delivery has much in common with the current web architecture underlying W3C activities, in that meta data is exchanged between end-devices and intermediaries to contextualize service delivery. Effective multi-device, multi-media delivery requires further standardization in the following areas:

W3C's CC/PP and IETF OPES activities provide effective building blocks to move forward from. CC/PP (and CC/PP protocol) extensions to support time-variant device attributes would allow the intermediary to keep track of dynamic changes to a device neighborhood. Similarly, extensions are needed to the CC/PP protocol  to convey capabilites of device bundles rather than single devices. Current work in OPES provides technical and business model standardization for the intermediary architecture. The current OPES architecture does not preclude a transcoder tailored for multi-device delivery. [MyVV] describes a multi-device, multimedia delivery testbed we have built with fairly minor extensions to CC/PP, and an OPES-compatible delivery framework.