混合式具路徑感知識別碼之分配與繞徑機制

Abstract

本篇論文提出一套混合式具路徑感知識別碼（Hybrid Path-aware Identity, HPID）的分配與繞徑機制，本機制將現有之具路徑感知識別碼(Path-aware Identity, PID)的分配機制加以整合，可針對不同網路拓樸，達到有效的識別碼分發。

PID係指識別碼本身隱含繞徑資訊，其編碼經運算後可以達到路徑選擇的目的。PID的使用不僅可以免除管理繞送路徑的繁複訊息，同時也可避免建立繞送路徑所產生的延遲。

然而各方所提之PID機制皆有某些限制與缺失，僅適用於部分網路環境。因此，本篇論文提出之混合式具路徑感知識別碼（Hybrid Path-aware Identity, HPID）乃針對不同實體分佈之需求，對識別碼加以規劃切割，並結合不同PID分配與繞徑機制，達到各機制間的特性整合。因此，HPID除了具有PID的特性及優點外，亦能根據不同實體分佈環境採用適性化的識別碼分配與繞徑機制。

根據模擬結果顯示，本機制可大幅提升各網路實體取得識別碼之成功率，進而縮短識別碼，降低訊息長度，同時減輕繞徑處理時之運算，達到省電之效用。

In this thesis, we propose a mechanism that can assign Path-aware IDentity (PID) much more efficiently by integrating two or more different PID assignment mechanisms based on the un-derlying network topologies. PIDs are identities that encode routing information into the val-ues of PIDs so that network entities can deduce routing paths from PIDs simply by performing mathematical operations on PIDs without exchanging routing messages. Therefore, PID as-signments not only can eliminate the maintenance overhead of routing tables but also reduce the routing delay.

However, conventional PID mechanisms have their own intrinsic limitations and cannot operate efficiently in various network environments. This thesis proposes a Hybrid Path-aware IDentity (HPID) assignment and routing mechanism that can combine two or more PID mechanisms in accordance of the underlying network topology. Based on the characteristics of the network environment, the HPID mechanism divides an identity into several partitions and adopts a specific PID assignment and routing mechanism for each partition. As a consequence, the HPID mechanism can inherit the salient features and advantages of different PID mecha-nisms, and thus can adapt to various network environments.

The simulation results show that the HPID mechanism significantly increases the success probability for an entity to acquire a unique identity. Furthermore, it also can shorten the iden-tity length and the computing overhead in routing path selection, and thus reduces the power consumption of each network entity.