SEG Algorithm

Abstract: Spectrum efficiency for a TDMA/FDMA microcellular system, where the base station does not load with full system channels as is seen in a conventional cellular system, can be heavily deteriorated if a coordinated use of the channels is not adopted: the random use of the channels on the same frequency between neighboring cells is most harmful. The authors show that the channel segregation (SEG) which has been proposed as a sophisticated dynamic channel assignment (DCA) method, has a capability to resolve this problem. Results of the analysis with computer simulation reveal that the SEG gives a higher spectrum efficiency compared with the conventional DCA, when applied to a TDMA/FDMA microcellular system. No significant modification to the SEG algorithm is required except that the channel usage is restricted depending on the channel usage situation due to the partial loading of the channels. >

Abstract: Trend and outlier are frequently used to derive early warning predictive signal to decision maker in order to achieve ultimate quality decision outcome in domain specific (e.g. commercial, scientific, biomedical and engineering, just to name a few) applications. We develop a gradient-based algorithm using sample entropy gradient(SEG) for trend and outlier prediction in high frequency time series data streams. L2 similarity measure (Euclidean distance between two linearized gradient curves is then computed and used to quantify the degree of similarity and compared with a threshold L2 value to judge the extend of dissimilarity that would be classified as outlier. SEG algorithm which circumvents the need to pre-specify tolerance parameter in those cross sample entropy (CSE)-based algorithms that invariably involve real domain expert to set the tolerance threshold. We conduct real data experiments on SEG algorithm to two application areas: dynamic wind speed data stream; and financial time series data. Our experiments demonstrated that SEG algorithm can be feasibly used in online implementation to derive predictive early warning signals to domain-specific decision maker.

Abstract: SAX (Symbolic Aggregate approXimation) is a kind of Symbolic time series similarity measurement method, which can not effectively distinguish the similarity between series in the circumstance of the corresponding value being similar between two sub-segment of time series. In this work, we proposed a novel time streams similarity approach based on SAX which was named KP_SAX. The similarity distance of KP_SAX described not only the statistical discipline of time series numerical change, but also the form changes of time series. The results show the superiority of our approaches as compared to the similarity measures of SAX and provide our promising results. Based on analysis above, an improved SAX algorithm based on Key Point, named as KP_SAX, is proposed in this paper. The algorithm is divided into two parts. Part 1, KP_SEG algorithm, figure out the Key Points of the time series. Based on the average segmentation point, Key Points are extra chosen as segmentation points, ensuring that the KP_SAX algorithm takes into full consideration the division according to intrinsic change law as well as the form change of the series, in order to efficiently differentiating the time series similarity. Part 2, KP_MEASURE algorithm, measure the similarity distance between two series based on the KP_SEG algorithm. According to the KP_SEG algorithm, the series are transformed into word sequences according to SAX symbolization delineation method, then converted to the similarity distance oftwo time series. Four time series ofgrain product from 1852 to 1925 of Rothamsted region is chosen in the experiment and the similarity is respectively measured by using SAX algorithm and KP_SAX algorithm. The experiment show that the result of the similarity of the four series are all zero when applying SAX algorithm, indicating SAX can't differentiate effectively. In comparison, KP_SAX algorithm can effectively measure the similarity distance between different time series, despite partly raised the time and space complexity ,achieving the goal of improving efficiency of SAX . II. SELECTION OF THE KEY POINTS IN TIME SERIES First ,we give the explanation of some symbols used in this paper: