自從封包無線網路出現，隨意網路就有很多協定被發展，這種網路是許多移動設備點的集合，經常是沒有中心控制及基本設施站，過去幾年這方面的研究快速成長，因為不昂貴的設備及可用的廣泛性。隨意網路所面對的挑戰包括移動性、能量受限、低頻寬及高錯誤率等。本論文提供無線網路綜觀及探討隨意無線網路與無線感測器網路的路由協定，並提出四種協定，分別用於不同網路拓僕架構，並說明四種協定之比較及適用場合，四種協定分別為 (一)具適應性與捷徑式的地理路由運作在隨意網路甚至網路具有空洞地帶(ASGR)，(二)運用模糊邏輯之多準則權重動態路由壽命方法達多重傳播路由之可靠性 (FMAR)，(三)區域化及負載平衡的群集方式的路由達無線感測網路之節能(LLBC)，(四)玫瑰方式開展之快速且均勻的路由規則運作於無線感測網路(ROSE)。
(一) ASGR(Adaptive Shortcut Geographic Routing)運用無需控制封包之貪婪式路由方式、具有無迥路、倒回原路、捷徑式、滿足有限記憶需求之特性，並包含路徑修護及保證封包送達路徑等功能。因此轉交點可有效率的轉送封包到達終點甚至遭遇到網路空洞地帶、運用抄捷徑之方法縮短至目的地的距離，使接近最短路徑，同時可運用於多重路徑。
(二) FMAR (Fuzzy Modified AODV Routing)是以模糊邏輯修正AODV方法的路由協定運用在隨意網路多重傳播路徑，動態路由壽命的評估在無線網路是一種挑戰，模糊邏輯多準則權重的方法用以動態評估活動中的路由壽命，從而決定最適路由。 FMAR建立多個路徑提供視訊或多重傳輸運用以加強穩定傳輸並克服傳輸之不可靠性與無線網路之有限頻寛。
(三) LLBC (localized load-balanced clustering) 的協定在於增長無線感測網路之壽命並能節省其能源，LLBC有兩種方法，一是ICHR(improved cluster head rotation) ，另一是MSC (modified static clustering) 。 ICHR用目前的群主點(cluster-head) 從該群中選擇最富能量的感測器點作為下一輪的群主點並避免重覆選擇邊緣的點為下一輪群主點。而MSC則適用在網路有少數甚富能量的感測器之情況，它只需執行少次的群主點輪選，因而可減少群主輪選之多餘耗損。它調整每個群集的成員數使儘可能接近平均數、以達群集負載平衡。
(四) ROSE是快速均化的路由協定，應用差集與相對度(Relative Degree, that is the cardinality of a set-head)等兩觀念來做感測器網路之路徑傳輸。此轉輸有如玫瑰開展，以樹狀架構將感測器點組成階層的集合(組)，由組頭(set-head)管理。起源點順著樹狀結構去找出所有點路徑，組頭執行資料熔解及聚合以減少傳輸時頻寛消耗、存取延遲、能量耗損，它保證路徑可讓封包送達，若組頭失敗，其組員變成孤兒，斷裂部分用上游復原方法很快黏回樹，以復原路徑。
本文提出四種protocols之比較及適應環境，建議可依不同環境狀況及目標需求，選取適用之routing protocol，以達成目標。將來可擴展ASGR、FMAR、LLBC 和 ROSE之加值運用 (例如增加網路安全)，或整合運用，可得更有效隨意無線網路路由。

Since the advent of packet radio network, numerous protocols have been developed for ad hoc network. The network is a collection of mobile nodes. Usually, it has no centralized or fixed infrastructure. Within past few years, the researches on this field grow rapidly because wireless devices are inexpensive and widely available. The challenges of ad hoc network include mobility, power limitation, narrow bandwidth and high error rates. The dissertation gives an overview of ad hoc network and wireless sensor networks and explores the routing protocols, (I) “Adaptive Shortcut Geographic Routing (ASGR) for Ad Hoc Networks Even with Topology Voids and Islands”, (II) “Fuzzy Logic Weighted Multi-Criteria of Dynamic Route Lifetime for Reliable Multicast Routings in Ad hoc Networks”(FMAR), (III) “Localized and Load-Balanced Clustering (LLBC) for Energy Saving in Wireless Sensor Networks”, and (IV) “Rose (ROSE) Algorithm for Fast and Even Routing in Sensor Networks”.
ASGR has the features of beaconless greedy forwarding, loop-freeing, backtracking, backward handling, short-cutting, limited memory requirement, route recovery algorithm and guaranteed packet delivery, so that a forwarding node can forward packets efficiently to the destination for topologies even with voids and islands. Since the short-cut rule is applied to shorten the path to the destination, the found short-cut path approaches the shortest path. ASGR can be also applied to multi-path, where sources can explore their own paths or take the advantage of the historical paths to reach the same destination.
FMAR is a fuzzy logic modified AODV routing protocol for multicast routing in ad hoc networks. Since the evaluation of dynamic route lifetime is challenging in the wireless networks, the fuzzy logic weighted multi-criteria of the protocol is used to dynamically evaluate the active route lifetime in order to determine the appropriate routes. In FMAR, multiple routing paths are established to provide extra schemas of video streaming or multicast transmission and enhance the robust transmission against unreliability and limited bandwidth of wireless links.
LLBC is a protocol for the energy savings and the lifetime increase of wireless sensor networks. LLBC contains two approaches. One is ICHR (improved cluster head rotation) and the other is MSC (modified static clustering). ICHR use the present cluster heads to select most energetic sensors as the next-round cluster heads and avoids the margin cluster heads being selected as cluster heads repeatedly. MSC is suitable when the network has a few very high energetic sensors. It only needs to perform a few cluster head rotations so the rotation overhead is reduced. It uses the method of inter-cluster load balance to adjust the cardinality of each cluster as close to the average cardinality as possible.
ROSE is proposed for fast and even routing in sensor networks. The algorithm uses the concepts of the differences of sets and the node relative degree to find the paths from the sink to all nodes in the network. It is developed in the manner of a rose blossoming. The protocol forms a tree-architecture to organize the sensors as hierarchical sets governed by their set-heads. Therefore, the sink can conveniently and quickly follows the tree structure to find its destinations. The set-heads perform data fusion / aggregation to reduce bandwidth consumption, media access delay and power consumption in communication. Rose Algorithm is proved to be packet delivery guaranteed. If a set-head failed, its children become orphan nodes. The broken fragments can be quickly glued back to the tree using the upstream local recovery method.
ASGR, FMAR, LLBC and ROSE could be extended (as increasing network security) or combined for very efficient routings in ad hoc wireless networks / wireless sensor networks.

[1] Murthy, C. S. R. and Manoj, B. S., “Ad Hoc Wireless Network Architectures and Protocols”, Pren