In the publish/subscribe paradigm, data providers disseminate publications to 
all consumers that have registered subscriptions. Most of the research in 
publish/subscribe systems is focused on semi-structured data where publications 
are represented as a set of (attribute, value) pairs and subscriptions express 
constraints over attribute values. In this paper we introduce a 
publish/subscribe paradigm for time series data. Publications and subscriptions 
in this model correspond to numerical sequences of fixed length and the 
semantics of matching is based on the notion of similarity between such 
sequences. Since time series can be modeled as higher dimensional points we 
evaluate the applicability of an existing spatial data indexing approach for the 
support of matching in such type of system. We provide a framework for 
developing time series publish/subscribe middleware which domain experts could 
easily extend to fit a particular application.