Bai toan thiet ke topo ao dugc phat bieu nhu sau: Dira ra mpt topo vat ly (so mit mang, so lien ket giua cac cap mit, so buoc song tren moi lien ket) va mpt ma tran luu lugng (traffic) yeu cau truyen thong (ket noi) trung binh giua eae cap mit nguon va dich trong mang.O day chiing toi gia su rang moi yeu
cau truyen thong giúa hai mit neu duoc thiet lap se
sir dung toan bp bang thong ciia mpt (va chi mpt) kenh quang dugc diing de thiet lap lightpath cho nọ Dieu nay co nghia rang, gia trj tuong ung ciia no trong ma tran luu lugng se bang 1 neu ton tai yeu truyen thong va bang 0 trong truong hgp ngugc laị Thiet ke mot topo ao tiíc la cai dat eae lightpath cho cac yeu cau truyen thong giúa cac cap mit nguon va dich nay sao cho tai nguyen mang duge su dung la toi uu (it nhat). Tai nguyen mang co the dugc hieu nhu la: so buoc song phai sií dung tren moi lien ket quang, so hop trung binh (so lan chuyen doi quang hay dp dai) ciia cac lightpath, thoi gian tre,... Trong do, so buoc song tren moi lien ket quang cua topo vat ly la tai nguyen quan trpng nhat [2].
1.2 Nhirng nghien ciru triroc
Bai toan thiet ke topo ao thuong dugc phat bieu duai dang quy hoach nguyen tuyen tinh hon tap
(mixed-integer linear programming) [1] va duoc
chiing minh la bai toan NP-day diị Vi vay cac thu^t toan heuristic la thuong dugc su dung de giai quyet no trong thue tẹ
Tac gia trong [2] da trinh bay thuat toan HLDA (heuristic logical topo design algorithm) voi muc dich cue tilu hoa so buoc song tren cac lien ket quang. Y tuong chinh cac thuat toan nay la su dung cac duong ng^n nhat giira cac cap nguon va dich de
thiet lap eae lightpath va do la diem yeu ciia thuat
toan naỵ D8 vugt qua cac diem yeu nay, cae tac gia trong [3] su dung thuat toan mo phong toi luyen
(simulated annealing) di thik k§ topo aọ Tuy nhien,
bai bao nay chi quan tam den mpt dang dac biet cua viec thilt ke topo ao do la hypercube topọ Cac tac
gia trong [1] chuyen tir bai toan thiet ke topo ao ve dang bai toan to mau do thi sau do ap dung cac thuat toan to mau do thj truyen thong. Viec chuyen tu bai toan thiet ke topo ao ve bai toan to mau do thi dugc thuc hien nhu sau:
1. Tim duong di ngan nhat cho eae cae cap nguon va dich. va dich.
2. Xay dung do thi phu bang each;
ạ Moi duong di tuong ung voi mot lightpath tao ra mot dinh
b. Hai lightpath co chung mpt lien ket vat ly (link) thi tao mpt cung giira hai dinh ly (link) thi tao mpt cung giira hai dinh tuong ling voi hai lightpath dọ
Viec cai dat cac lightpath bao gom hai phan: (1) tim duong di giua cac cap nguon va dich cho moi ket noi (2) an dinh cac buoc song cho eae duong di dọ Toi uu tai nguyen mang chinh la cue tieu so buoc song can su dung. Dieu nay tuong duong voi viec to mau do thi xay dung nhu tren su dung so mau it nhat. Trong bai toan to mau do thj hai dinh ke nhau khong dugc to cung mot mau, dieu nay tuong duong voi viec hai lightpath khac nhau co chung it nhat mpt lien ket vat ly phai sii dung hai buoc song khac nhaụ Do vay, bang each xay dung do thj nhu tren, bai toan thiet ke topo ao la tuong duong voi bai toan to mau do thj. Cach tiep can nay la tot nhung cung ton tai diem yeu giong voi cac thuat toan trinh bay trong [2] do la: o buoc 1 trong viec xay dung do thi ehiing ta da su dung duong di ngan nhat nhung cac tac gia trong [4] da chi ra rang: chi su dung duong di ngan nhat de cai dat cac lightpath khong dam bao toi uu tai nguyen dupe sii" dung.
De khac phuc nhugc diem ciia cac thuat toan trinh bay d tren (khong chi su dung duong di ngan nhat giija cac cap nguon va dich), mpt each tu nhien, chiing ta se thay la: phai xet cho tat ca cac duong di giira mpi cap nguon va dich. Gia su chiing ta co  yeu cau truyen thong va moi cap nguon dich tuong
ung <5„ dy> CO ^1 duong di giiia chiing (trong topo
vat ly). Nhu vay chiing ta phai ap dung thuat toan to mau do thj cho ATI x ^^ x ... x ^^. do thi co kich thuoc la Á" mit (chii y rang  la rat Ion trong thuc te). sau do chpn ra do thj co so mau can to la it nhat se la giai phap toi uu ciia bai toan. Ro rang dieu nay la rat kho thuc hien trong thuc tẹ Dua tren y tuong nay, tac gia trong [5] da dua ra thuat toan di truyin (genetic algorithm) cho viec thiet ke topo aọ Thuat toan nay khong chi cho phep giam khong gian tim kiem ma van hoi tu den cac giai phap toi uu toan cuẹ Mac du vay, toe dp hoi tu ciia thuat toan nay la
qua cham boi cac qua trinh lai ghep va dot bien trong qua trinh tiln hoạ D I kh^c phuc nhugc diem ciia qua trinh lai ghep va dot bien ciia cac thuat toan di truyen, cac tac gia trong [6] da dua ra mpt ky thuat toi uu hoa theo nhom bay (PSO-Particle
Swarm Optimization). Ky thuat nay cQng da duoc
chi ra sir hieu qua trong viec giai quyet eae bai toan dinh tuyIn trong mang cap quang [7]. Vi vay trong bai bao nay chiing toi se su dung ky thuat nay de thi§t ke cac topo aọ
U To chirc cua bai bao
PhSn con lai ciia bai bao dugc to chuc nhu saụ Muc II trinh bay Icy thuat toi uu hoa theo nhom bay cho viec thiet ke topo aọ Muc III trinh bay eae ket qua thirc nghiem. Chiing toi dua ra cac ket luan va de xuit cac huong phat trien trong muc TV.
11.TOI UU HOA THEO NHOM BAY CHO VIEC THIET KE TOPO AO THIET KE TOPO AO
n.l Toi vu hoa theo nhom bay
T6i uu hoa theo nhom bay la mpt ky thuat toi uu
hoa nglu nhien dua tren mot quan the dugc phat
trien bai Dr. Eberhart va Dr. Kennedy [6], phong theo hanh vi ciia cac bay chim hay cac dan cạ PSO CO nhieu sir tuong tu nhu ky thuat tinh toan tien hoa trong thuat toan di truyen (GA-Genetic algorithm) [8]. Cung giong nhu GA, PSO dugc khoi tao voi mpt quSn the ciia cac giai phap ngau nhien va tim ki^m giai phap toi uu bang viec cap nhat vj tri cac phSn tụ Tuy nhien, khong giong nhu GA, PSO khong CO cac thao tac tien hoa nhu la lai ghep (crossover) hay dot bien (mutation).