【Turkey】ICC 2006 在火雞國！

剛從美國回來不久的我，還是整天在看自助旅行的網站和報紙上各家旅行社的廣告！一邊看一邊想著我什麼時候能再出國去玩個一趟！計劃著趁三月寫完論文之後，就 來一趟印度或是柬埔寨之旅，然後八月去日本看世界盃籃球賽，十二月有幸的話，還能到卡達參加亞運會！不過，這些計劃都仍是在天馬行空的時候，在2006年 的第二天，我投到土耳其伊斯坦堡的論文結果回來了！

既 然寫了這麼一篇文章，那就表示是好消息！沒錯，的確是個很好的消息，我的論文不但被接受，而且還有再次上台做 Oral Presentation的機會，更好的是，日期就在這次會議的第二天！那表示，我報告完之後，還有好幾天的時間有著悠閒的心情在土耳其好好的遊覽遊覽 了！

土耳其是個什麼樣的國家？其實我也還搞不太懂！至少我在查機票的網站上知道，他有時候被分在歐洲、有時候被分在亞洲！他跟希臘相連，也跟敘利亞和伊朗相連！這麼講應該會嚇跑很多人吧，畢竟中東不是個很安全的地方！不過對我來說，卻是個有致命吸引力的地方！

ICC 的時間是六月11日到15日，我可以提早幾天出發，然後晚個幾天回來！這次不像美國之行有那麼長的假期了，我預計扣掉飛行，大概就十天左右的行程吧！打算 去的地方除了伊斯坦堡之外，還想去安卡拉、卡帕多奇亞、棉堡和愛琴海岸，時間可能不算多，但是我相信這一定會是一趟讓我很難忘的旅程！

現在就先跟大家說囉，我要去土耳其！要明信片的就快把住址給我吧！然後我再忙個論文到三月，就能真正動手排行程囉！

一句話，論文被接受，真是太爽了！離畢業又貼近了一大步！

轉貼個論文審查者（reviewer）的意見給大家看看吧！

REVIEWS:

===== Review =====

*** Comments To The Author: Please provide some comments to the author that explain your assessment

The paper proposes a novel fixed-alternate routing algorithm for optimal assignment of traffic intensities in all-optical networks so as to reduce the connection blocking probability. The authors have propsed a new objective function for finding the optimal way to split the traffic for a source-destination pair over multiple disjoint paths between the source-destination pair. The optimization problem uses the aggregate blocked traffic intensity as the objective function.

The convexity of the objective function is proved and hence the existence of a global optimum solution is estabished. The performance of the proposed algorithm is compared with fixed-alternate routing algorithm (that sorts the multiple paths acording to hop counts) and with the least-loaded dynamic-alternate routing algorithm. The proposed algorithm is observed to reduce significantly the connection blocking probability as compared with hop-count based fixed-alternate routing mechanism. The call blocking probability of the proposed algorithm is very close to the least loaded dynamic-alternate routing mechanism.

Simulations also show that the proposed algorithm also achieves better fairness than the other fixed-alternate routing mechanism.

The paper is very well written. The problem is motivated clearly, with a good account of prior works. By intelligently casting the objective function in terms of blocked traffic intensity, the optimal solution is obtained.


===== TPC Review =====

*** Comments To The Author: Please provide some comments to the author that explain your assessment

The authors provide very clear formulation as the constrained convex optimization problem that makes this paper theoretically sounds.

===== TPC Review =====

*** Comments To The Author: Please provide some comments to the author that explain your assessment

This paper proposed a new fixed-alternate routing algorithm for all-optical WDN networks. The basic idea of their algorithm is to sort candidate routes between a pair of s-d nodes according to the amounts of traffics routed through the paths in the optimal solution. Given the difficulty of calculating blocking probabilites, the authors uses the approximate amount of blocked traffic intensity as the objective function, which has been proved to be convex.

The simulations are conducted on randomly generated graphs, and the average blocking probabilites along with the standard deviations over these random graphs are presented. They have shown that the performance of the proposed TI-FAR is very close to or even slightly better than the performance of a dynamic routing algorithm: LLR.

The reviewer has one concern: the input to the problem is the traffic matrix, which is assumed to be Poisson process with constant arrival rate; and the holding time is also exp. distributed with constant mean. In case the parameters of the Poisson process and/or holding time are changed, it may need to recalculate the importance of each alternate routes, and the order of the alternate routes may change. In another word, the TI-FAR algorithm still relies on the traffic pattern, which may be changing all the time. In the sense, the TI-FAR may not be called "fixed-alternate routing".