TY - GEN
T1 - Load balancing regenerating codes for multimedia content streaming
AU - Corena, Juan Camilo
AU - Ohtsuki, Tomoaki
PY - 2013/12/1
Y1 - 2013/12/1
N2 - In this article, we explore the use of a special type of erasure codes known as Regenerating Codes (RCs), as a way to perform load balancing among servers of a multimedia content streaming site. The goal of our construction is to reduce the costs of keeping the redundant servers by minimizing the amount of information that must be stored on disks as well as main memory. This proposal is advantageous for devices connecting through several means, such as a specialized video streaming device with two different internet connections or a smartphone connecting through Wi-Fi and a carrier simultaneously. To achieve this goal, we use a regenerating code based on a technique introduced by Rashmi et al. known as the Twin Code Framework. The use of this technique allows any user to connect to any k servers containing encoded fragments of the file to decode a particular part of it, while keeping the information overhead at the servers low. The construction also allows new copies to be created when demand increases; copies created at a later stage can be used in conjunction with previously created copies for load balancing purposes. In addition, any node in the system can be repaired with a relatively low amount of transmitted information compared to the length of the stored contents. Our instantiation of the framework uses Network Coding with coefficients drawn from a Hilbert Matrix.
AB - In this article, we explore the use of a special type of erasure codes known as Regenerating Codes (RCs), as a way to perform load balancing among servers of a multimedia content streaming site. The goal of our construction is to reduce the costs of keeping the redundant servers by minimizing the amount of information that must be stored on disks as well as main memory. This proposal is advantageous for devices connecting through several means, such as a specialized video streaming device with two different internet connections or a smartphone connecting through Wi-Fi and a carrier simultaneously. To achieve this goal, we use a regenerating code based on a technique introduced by Rashmi et al. known as the Twin Code Framework. The use of this technique allows any user to connect to any k servers containing encoded fragments of the file to decode a particular part of it, while keeping the information overhead at the servers low. The construction also allows new copies to be created when demand increases; copies created at a later stage can be used in conjunction with previously created copies for load balancing purposes. In addition, any node in the system can be repaired with a relatively low amount of transmitted information compared to the length of the stored contents. Our instantiation of the framework uses Network Coding with coefficients drawn from a Hilbert Matrix.
UR - http://www.scopus.com/inward/record.url?scp=84893285634&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84893285634&partnerID=8YFLogxK
U2 - 10.1109/PIMRC.2013.6666766
DO - 10.1109/PIMRC.2013.6666766
M3 - Conference contribution
AN - SCOPUS:84893285634
SN - 9781467362351
T3 - IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, PIMRC
SP - 3558
EP - 3562
BT - 2013 IEEE 24th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications, PIMRC 2013
T2 - 2013 IEEE 24th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications, PIMRC 2013
Y2 - 8 September 2013 through 11 September 2013
ER -