TY - GEN
T1 - A proposal of a quasi-solution state evolution algorithm for channel assignment problems
AU - Funabiki, Nobuo
AU - Nakanishi, Toru
AU - Yokohira, Tokumi
AU - Tajima, Shigeto
AU - Higashino, Teruo
N1 - Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 2002.
PY - 2002
Y1 - 2002
N2 - The channel assignment problem (CAP) in a cellular network requires finding a channel assignment to the call requests from cells such that three types of interference constraints are not only satisfied, but also the number of channels (channel span) is minimized. This paper presents a three-stage iterative algorithm, called the Quasi-solution state evolution algorithm for CAP (QCAP). QCAP evolutes quasi-solution states where a subset of call requests is assigned channels and no more request can be satisfied without violating the constraint. The first stage computes the lower bound on the channel span. After the second stage greedily generates an initial quasi-solution state, the third stage evolutes them for a feasible solution by iteratively generating best neighborhoods, with help of the dynamic state jump and the gradual span expansion for global convergence. The performance is evaluated through solving benchmark instances in literature, where QCAP always finds the optimum or near-optimum solution in very short time.
AB - The channel assignment problem (CAP) in a cellular network requires finding a channel assignment to the call requests from cells such that three types of interference constraints are not only satisfied, but also the number of channels (channel span) is minimized. This paper presents a three-stage iterative algorithm, called the Quasi-solution state evolution algorithm for CAP (QCAP). QCAP evolutes quasi-solution states where a subset of call requests is assigned channels and no more request can be satisfied without violating the constraint. The first stage computes the lower bound on the channel span. After the second stage greedily generates an initial quasi-solution state, the third stage evolutes them for a feasible solution by iteratively generating best neighborhoods, with help of the dynamic state jump and the gradual span expansion for global convergence. The performance is evaluated through solving benchmark instances in literature, where QCAP always finds the optimum or near-optimum solution in very short time.
UR - http://www.scopus.com/inward/record.url?scp=84937434967&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84937434967&partnerID=8YFLogxK
U2 - 10.1007/3-540-45801-8_4
DO - 10.1007/3-540-45801-8_4
M3 - Conference contribution
AN - SCOPUS:84937434967
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 32
EP - 41
BT - Information Networking
A2 - Chong, Ilyoung
PB - Springer Verlag
T2 - International Conference on Information Networking, ICOIN 2002
Y2 - 30 January 2002 through 1 February 2002
ER -