Data-stream-based global event monitoring using pairwise interactions

Abstract

The problem of global state observation is fundamental to distributed systems and to the analysis of data streams. Many interactions in distributed systems can be analyzed in terms of the building block formed by the pairwise interactions of intervals at two processes. Considering causality-based pairwise interactions by which two processes may interact with each other, there are 40 orthogonal interaction types. For each pair of processes (P i , P j ), let interaction type r i, j be of interest. This paper examines the problem: "If a global state of interest to an application is specified in terms of such pairwise interaction types, one per pair of processes, how can such a global state be detected?" A solution identifies a global state in which the interaction type specified for each process pair is satisfied. This paper formulates the specific conditions on the communication structures to determine which of the intervals being examined at any time may never satisfy the stipulated interaction type for that pair of processes, and therefore that interval(s) need no longer be considered as forming a part of any solution. Based on this theory, the paper proposes two on-line distributed algorithms to solve the problem.

Repeated detection of conjunctive predicates in distributed executions

“…This was formalized as the problem Fine_Rel in [2] and a O (n 2 p) time algorithm was given to detect the solution. The theory was further extended and distributed algorithms were given in [3] to solve this problem. Polynomial time solutions were possible only under a certain condition that was specified using the prohibition function.…”

“…We can observe that for the Fine_Rel modalities, the CONVEX-ITY property will not hold for detecting all solutions in polynomial time. Once a solution set I is detected (using the algorithms in [2,3]), we need to be able to safely prune at least one of the intervals in I to avoid queue build-up, analogous to the first challenge in Section 3. Define R I ij (X i , Y j ) ∈ r * ij to be the relation from that holds between intervals X i , Y j ∈ I.…”

Data Alignment for Data Fusion in Wireless Multimedia Sensor Networks Based on M2M

“…To achieve this, it is necessary to detect patterns, such as concurrency, among the generated streams. As has been shown in the works of Chandra and Kshemkalyani [3], [18], [19], a practical way to detect patterns among local-streams is by assuming a global time axis and by using the interval-interval relations defined by Allen [16]. Unfortunately, establishing a global timeline in a WMSN is difficult due to the lack of perfectly synchronized clocks [4].…”

Execution and Time Models for Pervasive Sensor Networks

“…The P ossibly and Def initely modalities [10] have been the most widely used. Refining these further, a complete suite of 40 orthogonal relationships among time intervals at two different physical locations (see [7,8,20,21]) was used to specify causality-based relationships among the local values that held during the local time intervals. Then, given a system with n processes, a specification space of size (2 40 −1)C n 2 for fine-grained relationships was identified.…”