xinchen liu - Academia.edu (original) (raw)
Related Authors
Università Cattolica del Sacro Cuore (Catholic University of the Sacred Heart)
Università Cattolica del Sacro Cuore (Catholic University of the Sacred Heart)
Uploads
Papers by xinchen liu
2016 IEEE International Conference on Multimedia and Expo (ICME), 2016
Proceedings. Eighth International Conference on Parallel and Distributed Systems. ICPADS 2001, 2001
Wavelength-division multiplexing (W D M) optical networks provide huge bandwidth by allowing mult... more Wavelength-division multiplexing (W D M) optical networks provide huge bandwidth by allowing multiple data streams transmitted simultaneously along the same optical fiber, with each stream assigned a distinct wavelength. A key issue o n W D M optical networks is to minimize the number of wavelengths f o r realizing a routing request. Let W be the number of wavelengths supported by a W D M optical network. For a routing request R which needs 1 wavelengths, i f I 5 W then R can be realized in one round of routing. However, when 1 > W , multiple rounds of routing f o r R are required. I n this case, it is important t o minimize the number of routing rounds. Multicast is to transmit a data stream f r o m one input t o multiple outputs (one-to-many), a fundamental communication pattern in many applications. I n this paper, we study the problem of minimizing the number of wavelengths and the number of routing rounds f o r realizing a set R = { (u , v) } of multicasts, where each output v receives a data stream f r o m exactly one input U , o n the n-dimensional W D M all-optical multistage interconnection networks (MINs). For the network with wavelength converters, we show that any set of multicasts can be realized by 2 r (n-1) f (k + ') 1 wavelengths in k rounds of routing. For one round of routing, the upper bound 2r(n-1)f21 is tight to the lower bound. W e also give algorithms f o r multicasts o n the network without wavelength converters. Computer simulation results show that any set of multicasts can be realized in at most two rounds of routing o n the network with practical size.
2016 4th International Conference on Cloud Computing and Intelligence Systems (CCIS), 2016
In current cities, the number of vehicles grows rapidly es- pecially in developing countries, and... more In current cities, the number of vehicles grows rapidly es- pecially in developing countries, and the trac surveillance system usually has tens of thousands of cameras connected into a huge network. Hence the volume of data generated by trac cameras becomes astronomical. So it is a great chal- lenge to process and utilize these big data resources e ec- tively and eciently. Towards this end, this paper proposes a system which provides a novel service of vehicle trajectory search for urban trac surveillance. In this system, smart cameras extract vehicle IDs with time information when ve- hicles appear in their views and send these information to a data center with very little bandwidth cost. After that, the center server stores and organizes trac data using two types of tables, camera tables and inverted tables. We fuse vehicle IDs, spatial-temporal data, and topology of urban roads to build a global graph and propose a PathRank algo- rithm to support the vehicle trajectory searc...
2016 IEEE International Conference on Multimedia and Expo (ICME), 2016
Proceedings. Eighth International Conference on Parallel and Distributed Systems. ICPADS 2001, 2001
Wavelength-division multiplexing (W D M) optical networks provide huge bandwidth by allowing mult... more Wavelength-division multiplexing (W D M) optical networks provide huge bandwidth by allowing multiple data streams transmitted simultaneously along the same optical fiber, with each stream assigned a distinct wavelength. A key issue o n W D M optical networks is to minimize the number of wavelengths f o r realizing a routing request. Let W be the number of wavelengths supported by a W D M optical network. For a routing request R which needs 1 wavelengths, i f I 5 W then R can be realized in one round of routing. However, when 1 > W , multiple rounds of routing f o r R are required. I n this case, it is important t o minimize the number of routing rounds. Multicast is to transmit a data stream f r o m one input t o multiple outputs (one-to-many), a fundamental communication pattern in many applications. I n this paper, we study the problem of minimizing the number of wavelengths and the number of routing rounds f o r realizing a set R = { (u , v) } of multicasts, where each output v receives a data stream f r o m exactly one input U , o n the n-dimensional W D M all-optical multistage interconnection networks (MINs). For the network with wavelength converters, we show that any set of multicasts can be realized by 2 r (n-1) f (k + ') 1 wavelengths in k rounds of routing. For one round of routing, the upper bound 2r(n-1)f21 is tight to the lower bound. W e also give algorithms f o r multicasts o n the network without wavelength converters. Computer simulation results show that any set of multicasts can be realized in at most two rounds of routing o n the network with practical size.
2016 4th International Conference on Cloud Computing and Intelligence Systems (CCIS), 2016
In current cities, the number of vehicles grows rapidly es- pecially in developing countries, and... more In current cities, the number of vehicles grows rapidly es- pecially in developing countries, and the trac surveillance system usually has tens of thousands of cameras connected into a huge network. Hence the volume of data generated by trac cameras becomes astronomical. So it is a great chal- lenge to process and utilize these big data resources e ec- tively and eciently. Towards this end, this paper proposes a system which provides a novel service of vehicle trajectory search for urban trac surveillance. In this system, smart cameras extract vehicle IDs with time information when ve- hicles appear in their views and send these information to a data center with very little bandwidth cost. After that, the center server stores and organizes trac data using two types of tables, camera tables and inverted tables. We fuse vehicle IDs, spatial-temporal data, and topology of urban roads to build a global graph and propose a PathRank algo- rithm to support the vehicle trajectory searc...