TY - GEN
T1 - Approximately optimal utility maximization
AU - Nedić, Angelia
AU - Subramanian, Vijay G.
PY - 2009
Y1 - 2009
N2 - All opportunistic scheduling algorithms solve simpler optimization problems at each scheduling instance in order to achieve good long-term performance. The analysis of these algorithms assumes that the simpler optimization problems are solved exactly. However, in contrast, real-life implementations only approximately solve these problems but still yield close to optimal performance. We formalize this observation by explicitly bounding the longterm performance in terms of the error in the approximation made at every stage.
AB - All opportunistic scheduling algorithms solve simpler optimization problems at each scheduling instance in order to achieve good long-term performance. The analysis of these algorithms assumes that the simpler optimization problems are solved exactly. However, in contrast, real-life implementations only approximately solve these problems but still yield close to optimal performance. We formalize this observation by explicitly bounding the longterm performance in terms of the error in the approximation made at every stage.
UR - http://www.scopus.com/inward/record.url?scp=77950671552&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=77950671552&partnerID=8YFLogxK
U2 - 10.1109/ITWNIT.2009.5158572
DO - 10.1109/ITWNIT.2009.5158572
M3 - Conference contribution
AN - SCOPUS:77950671552
SN - 9781424445363
T3 - Proceedings - 2009 IEEE Information Theory Workshop on Networking and Information Theory, ITW 2009
SP - 206
EP - 210
BT - Proceedings - 2009 IEEE Information Theory Workshop on Networking and Information Theory, ITW 2009
T2 - 2009 IEEE Information Theory Workshop on Networking and Information Theory, ITW 2009
Y2 - 10 June 2009 through 12 June 2009
ER -