QoC-based distribution

Quality of Context (QoC) usually applies to the quality of the distributed context data. Instead, SALES uses this concept to adapt context data distribution and delivery, so to save previous system-level resources. To guide management and reconfiguration operations, we introduce special contracts, i.e., Context Data Distribution Level Agreements (CDDLAs), to detail the quality level the infrastructure has to ensure.

Each CDDLA comprises three main objectives:

Freshness - Up-to-dateness requirement on matching data
Data retrieval time - The time needed by the mobile node to retrieve context data
Priority - Traditional priority value useful to enable traffic differentiation and to favor the routing of high-priority data under load conditions

For the clarity sake, we introduce three principal users’ classes that define associated quality parameters:

Gold [latest version of data, retrieval time 2s, priority 0]
Silver [valid data, retrieval time 4s, priority 1]
Bronze [non-valid data expired at most by 2 seconds, retrieval time 6s, priority 2]

At run-time, CDDLAs are used to derive the query parameters for data routing (for more details, see the associated literature). In particular, from a general point-of-view, it is possible to highlight the subsequent dependencies between CDDLAs and query parameters:

Query Parameters	Depends on	Details
Query Lifetime	CDDLA data retrieval time	Data routed with more than CDDLA data retrieval time are useless. Remove query to avoid unneeded overhead.
Query Priority	CDDLA priority	Use the same priority between query and CDDLA.
Horizontal Time To Live	CDDLA freshness	The broader the CDDLA freshness, the higher the probability of retrieving matching data on close peers. Hence, use larger horizontal time to live according to data freshness.
Query Routing Delay	CDDLA data retrieval time	QRD represents the maximum delay each node can apply during query distribution. The higher this delay, the higher the data retrievel time.
Data Routing Delay	CDDLA data retrieval time	DRD represents the maximum delay each node can apply during data distribution. The higher this delay, the higher the data retrievel time.