1.

Introduction

MANETs do not require fixed infrastructure, because they are selfconfigured wireless networks. Their nodes create connections with distant communication partners when is necessary. On the other hand, the radio transmission range is small, therefore, communication partners are often not within direct radio range and connections should be set-up over multiple nodes, being used these nodes as relays to forwarding data. Moreover, the network topology constantly changes because of node mobility. These changes cause frequent route breaks and force sources to re-establish or maintain connections to their distant communication partners. Numerous ad hoc routing algorithms exist to allow networking under various conditions. They can be traditionally separated into two groups, proactive and reactive algorithms. On the one side, proactive routing algorithms maintain continuously updated state of the network and the existing routes; however, in some cases it may generate an unnecessary overhead to maintain the routing tables and then may be better to create routes only on demand, the case of reactive routing algorithms. On the other hand, reactive routing algorithms require timeconsuming route creations that may delay the actual transmission of the data when sources have no path towards their destination and then, in this case may be better to use a proactive routing algorithm. Due to advantages and disadvantages of each kind of routing algorithm, we consider that the use of a hybrid routing algorithm may be suitable in different cases, having as big objective to route for real-time data and multimedia communication.

2. Routing in Mobile Ad Hoc Networks:

2.1 Concept Illustration

MANETs consist of a collection of wireless nodes, all of which can be mobile, that dynamically create a wireless network between them selves without a fixed infrastructure or administrative support. They can be created and used anywhere, any time offering unique advantages and versatility for certain environments and certain applications, i.e. for rescue agencies, policies, transporting companies or merely for office use. In the case of rescue problems, for example in data analysis in scenes of natural disasters, an ad hoc network could be formed by communication devices in fire brigades, helicopters, ambulances, policies and also people with laptop computers or mobile phones in hospitals, pharmacies and so on, all together working in a collaborative way to provide effective solutions to the problem. In figure 1, we show an example of an ad hoc network which has different communication devices and some connections amongst themselves (it is when devices are in of range between each other).

It is obvious that they need to communicate each other; however the high degree of mobility in this kind of network does networks change quickly. Some devices could be out of range with respect to others and therefore each device (node) must be able to act as a router to relay packets generated by other nodes. In figure 1, when node G needs to communicate to node I, node H has to act asa router and transmit its information. But, what happen when node H is out of range with respect to node I? It can exist the possibility that node E is now in of range and the topology of the network has changed to one different as it is shown in Figure 2. In this new topology, node G can communicate to node I through node D, E acting as routers.

Then, we can see what difficult the tasks of routing and maintaining paths in MANETs are and how much they are affected because of the node mobility, transmission signal, limitations on the battery power of mobile nodes, and moreover limitations on the resources in terms of bandwidth of the wireless medium and the fact of sharing this medium (most commonly controlled by the IEEE 802.11 Medium Access Control (MAC) protocol).

2.2 Existing Protocols

Routing in MANETs has traditionally used the knowledge of the

Figure 3: Routing in a network instantaneous connectivity of the network with emphasis on the state of the links [2]. This is the so-called topology based approach [3]. To overcome the problems associated with the link-state and distance vector algorithms, numerous routing protocols have been proposed. These protocols can be traditionally separated into two groups: periodic (also called proactive, global or table driven), and on-demand (also called reactive) even though there is a new generation of hybrid protocols which are both proactive and reactive in nature. Proactive protocols (Table driven) maintain a continuously updated state of the network and there with nodes are able to create instant connections to other nodes. In case of frequent topology changes, the necessary overhead to maintain the necessary link

tables often exceeds the advantage of quick route creations. Frequent routing packets overload the network and delay data packets or even cause packet drops. If nodes increase the period between consecutive topology updates, connectivity information in nodes possibly contains inaccuracies which can reduce network performance. Examples of proactive protocols are DSDV (Destination-Sequenced Distance-Vector Routing Protocol) [5], WRP (Wireless Routing Protocol) [6] and so on. The main differences among them are the number of used tables, the information that is kept and the forward packet police to maintain the tables updated. Reactive protocols (On demand) create and maintain routes only on demand and do not try to maintain an overview over the network. This means that routes are determined and maintained for nodes that require sending data to a particular destination. This reduces the generated overhead (typical in proactive protocols), but requires time-consuming route creations, as sources do not have any path towards their destination. Route discovery usually occurs by flooding a route request packets through the network. Examples of reactive protocols are AODV (Ad Hoc On-Demand Distance Vector Routing) [7], DSR (Dynamic Source Routing) [8] and so on. Reactive protocols also can be classified into two categories, source routing where each data packet carries the complete source to destination address and hop-by-hop routing where each data packet only carries the destination address and the next hop address.

Table-Driven vs. On-Demand Routing

Parameters Availability of On-Demand Routing Available when needed Flat Not Required Using localized Table-Driven Always available regardless of need Mostly Flat except for CSGR Yes route Inform other nodes to achieve

Information Routing Philosophy Periodic Route Updates Coping with mobility Signaling traffic generated

discovery as in ABR & SSR consistent routing table Grows with increasing Greater than that of Onmobility of active routes (as in Demand Routing ABR) Few can support QoS

Quality of service support

Mainly Shortest Path as QoS

metric Table: Overall Comparisons of On-Demand versus Table-Driven Based Routing Protocols.

Hybrid protocols try to profit the advantages of both reactive and proactive protocols and combine their basic properties into one. These protocols have the potential to provide higher scalability than pure reactive or proactive protocols thanks to the collaboration between nodes with close proximity to work together and therefore reduce the route discovery overhead. Examples of hybrid protocols are ZRP (Zone Routing Protocol) [9], DDR (Distributed Dynamic Routing) [10] and so on. Furthermore, each group of protocols has a number of different routing strategies, which employ a flat or a hierarchical routing structure.

3. Position Based Routing:

While in wireless networks with infrastructure support a base station always reaches all mobile nodes, this is not the case in an ad hoc network. Thus, routing is needed to find a path between source and destination and to forward packets appropriately. In Traditional Routing Algorithms like AODV [1], DSR [2], DSDV [3], a node has to disclose its ID in the network for building a router. A node activity such as sending or receiving data is highly traceable. Nodes are vulnerable to attacks and disruptions. Routing schemes rely on the cooperation and information exchanged among the nodes. More recently, there has been a growing focus on a class of routing algorithms that rely largely or completely on location information (based on position). These algorithms are commonly referred to as
POSITION-BASED ROUTING ALGORITHMS. Here in addition to node ID, extra information such as positions of the

nodes is used for making routing decisions. Since it is unlikely for 2 ad hoc nodes to stay at the same position concurrently, the match between a position and ID is definitely unique. Hence, in these algorithms, when positions are revealed for routing , there is no need of node IDs. Hence node anonymity can be maintained. However such algorithms rely on position exchange among the neighboring nodes. Such time based position exchange messages make a node highly traceable. The trajectory of a node movement can be well known to other nodes even when its node ID is intentionally hidden. Hence there is lack of privacy in traditional position based ad hoc routing algorithms. Position based routing algorithms claim that no routing tables need to be maintained and thus no overhead due to route discovery and route maintenance is imposed. But they need to obtain position data of their corresponding destinations, either by an internal discovery process, or by an independent position service, which will then impose overhead to maintain the position information (either proactively or on-demand).

Further in this paper, position based routing algorithms are compared in terms of their characteristics and forwarding strategies. Greedy algorithms like GPSR use either a most forward within radius or a nearest with forward progress strategy. It is argued, that NFP is of great advantage if the transmission radius/power can be controlled, and additionally have the benefit of reduced channel competition. Recent studies have shown that power control may not improve channel utilization much, because the longer path lengths (in hops) make up for the benefit. Greedy algorithms can route to a local maximum and need a recovery strategy in this case. Among several suggestions, the planar graph traversal methods seem to be the most reasonable. DREAM and LAR use a flooding approach, but packets are not sent to all neighbors, but only to those in the right direction of the target (i.e. the packet is forwarded to any node within the request zone). Finally some protocols use a hybrid/hierarchical scheme, like Terminode Routing. For long distances a greedy directional routing scheme is used. If the packet is close enough, some non-directional mechanism will guide the packet to the destination. In an ad hoc network, a message sent by a node reaches all its neighboring nodes that are located at distances up to the transmission radius. Because of the limited transmission radius, the routes between nodes are normally created through several hops in such multihop wireless networks. In the widely accepted unit graph model, two nodes, A and B, in the network are neighbors if the distance between them is at most R, where R is the transmission radius that is equal for all nodes in the network. Variations of this model include unit graphs with obstacles (or sub graphs of unit graphs), and manpower graphs where each node has its own transmission radius and links are allowed only when bi-directional communication is possible. No credible research was done in literature on any other model other than the unit graph model (one important exception is in [1]). However, in power and cost savings and congestion-aware methods, nodes may adjust their transmission power to merely reach an intended receiver. The use of the nodes position for routing poses evident problems in terms of reliability. The accuracy of the destinations position is an important problem to consider. In some cases the destination is a fixed node (e.g., a monitoring center known to all nodes, or the geographic area monitored), and some networks are static. We shall describe the following simple strategy. If a message is reasonably short, it can be broadcast (i.e., flooded) using an

optimal broadcasting scheme. If a message is relatively long, destination search (or route discovery) can be initiated, which is a task of broadcasting a short search message. The destination then reports back to the source by routing a short message containing its position. The source is then able to route the full message toward the accurate position of

the destination.

In this article we consider the routing task, in which a message is to be sent from a source node to a destination node in a given wireless network. The task of finding and maintaining routes in sensor and ad hoc networks is nontrivial since host mobility and changes in node activity cause frequent unpredictable topological changes. The destination node is known and addressed by means of its location. Routing is performed

by a scheme based on this information, generally classified as a position-based scheme. Table 1 contains a quick synopsis/superset of the schemes discussed in this article.

3.1 Advantages of using position based routing desitions;

The distance between neighboring nodes can be estimated on the basis of incoming signal strengths or time delays in direct communications. Relative coordinates of neighboring nodes can be obtained by exchanging such information between neighbors. Alternatively, the location of nodes may be available directly by communicating with a satellite (for outdoor networks), using GPS (Global Positioning System), if nodes are equipped with a small low power GPS receiver. The position-based approach in routing becomes practical due to the rapidly developing software and hardware solutions for determining absolute or relative positions of nodes in indoor/outdoor ad hoc networks. The routing algorithms should perform well for wireless networks with an arbitrary number of nodes. Sensor and rooftop networks, for instance, have hundreds or thousands of nodes. While other characteristics of each algorithm are easily detected, scalability is sometimes judgmental and/or dependent on the performance evaluation outcome. A scalable solution is one that performs well in a large network. It has been experimentally confirmed that routing protocols that do not use geographic location in the routing decisions, such as AODV, DSDV, or DSR are not scalable. For instance, describes GLS (scalable location service), similar to the doubling circle method independently proposed by Amouris, Papavassiliou, and Lu in 1999. Experiments using the ns simulator for up to 600 mobile nodes show that the storage and bandwidth requirements of GLS grow slowly with the size of the network. Furthermore, GLS tolerates node failures well: query performance degrades gracefully as nodes fail and restart, and is relatively insensitive to node speeds. Simple geographic forwarding combined with GLS compares favorably with DSR; in large networks (over 200 nodes), it

10

delivers more packets and consumes fewer network resources . Similar conclusions were made where the depth-first search-based GRA scheme was compared with the DSDV protocol. Routing table sizes in GRA were logarithmic vs. linear for DSDV (e.g., vs. 12 1000 in networks with 1024 nodes) . Therefore, it is likely that only position- based approaches provide satisfactory performance for large networks.

We shall now elaborate on other properties and reasons for difference in scalability. Localized algorithms are distributed in nature and resemble greedy algorithms, where simple local behavior achieves a desired global objective. In a localized routing algorithm, each node makes a decision to which neighbor to forward the message based solely on the location of itself, its neighboring nodes, and the destination. In shortest- (weighted)-path-based nonlocalized algorithms, each node maintains accurate topology of the whole network. In addition, since nodes change between active and sleep periods, the activity status for each node is also required. Although routing-table- (typical nonposition)- based solutions merely keep the best neighbor information on a route toward the destination, the communication overhead for maintenance of routing tables due to node mobility and topology changes is quadratic in network size (each change in edge or node status may trigger routing table modifications in a large portion of the network). On the other hand, position-based localized algorithms avoid that overhead, by requiring only accurate neighborhood information and a rough idea of the position of the destination. For example, edge and node changes in one part of the network have no immediate impact on almost any route. Clearly, only localized algorithms provide scalable solutions, especially for networks with critical power-constrained resources at nodes (e.g., sensor networks).

11

3.2 Qualitative characteristics of Position based routing algorithms: a)Loop-freedom. The proposed routing protocols should be inherently loop-free, to avoid timeout or memorizing past traffic as cumbersome exit strategies. Proposed algorithms are therefore classified as having or not having loop free property. b) Distributed operation. Localized algorithms [EGHK] are distributed algorithms that resemble greedy algorithms, where simple local behavior achieves a desired global objective. In a localized routing algorithm, each node makes decision to which neighbor to forward the message based solely on the location of itself, its neighboring nodes, and destination. Non-localized algorithms can be classified as global or zonal ones. In a global routing algorithm, each node is assumed to know the position of every other node in the network. In addition, since nodes change between active and sleep periods, the activity status for each node is also required. When such global knowledge is available, the routing task becomes equivalent to the shortest path problem, if hop count is used as main performance metrics (such an algorithm is described in [BCS, SWR]). If power or cost metrics are used instead, the shortest weighted path algorithm may be applied, as described in [RM] for power and in [SWR] for cost metric. Between the two extremes is the zonal approach, where network is divided into zones, with localized algorithm applied within each zone, and shortest path or other scheme applied for routing between zones [JL, LAR]. Clearly, localized algorithms are preferred if they can nearly match the performance of non-localized ones. An expanded locality is sometimes considered. For example, if two hop neighbors are included, the algorithm is classified as 2-localized. 12

c) Path strategy. The shortest path route is an example of a single path strategy, where one copy of the message is in the network at any time. Arguably, the ideal localized algorithm should follow a single path. On the other extreme are flooding based approaches, where message is flooded through the whole network area (broadcasting solves routing, and in high mobility scenario this could be optimal solution [HOTV], if optimized [P, QVL, SSZ]), or portion of the area [BCSW, KV]. The compromise is multi-path strategy, that is route composed of few single recognizable paths. Some algorithms are combinations of two strategies, and are appropriately labeled (e.g. single-path/flooding, single-path/multipath). d) Metrics. The metrics that is used in simulations normally reflects the goal of designed algorithm, and is naturally decisive in the route selection. Most routing schemes use hop count as the metrics, where hop count is the number of transmissions on a route from a source to destination. This choice of metric agrees with the assumption that nodes cannot adjust (that is, reduce) their transmission radii in order to reach desired neighbor with minimal power. It also assumes that delay is proportional to hop count (when the impact of congestion is not significant), and that the (both energy and bandwidth) cost of starting communication with neighbor is considerable (this is supported by the analysis in [Fe, FN]). However, if nodes can adjust their transmission power (knowing the location of their neighbors) then the constant metric can be replaced by a power metric that depends on distance between nodes [E, RM, HCB]. The goal is to minimize the energy 4 required per each routing task. However, some nodes participate in routing packets for many source-destination pairs, and the increased energy consumption may result in their failure. Thus pure power consumption metric may be misguided in the long term, and longer path that passes through nodes that have plenty of energy may be a better solution. The cost metric (a rapidly increasing function of decreasing remaining

13

energy at node) is used with the goal of maximizing the number of routing tasks that network can perform.

e) Memorization. Solutions that require nodes to memorize route or past traffic are sensitive to node queue size, changes in node activity and node mobility while routing is ongoing (e.g. monitoring environment). It is better to avoid memorizing past traffic at any node, if possible. However, the need to memorize past traffic is not necessarily a demand for significant new resources in the network for several reasons. First, a lot of memory space is available on tiny chips. Next, the memorization of past traffic is needed for short period of time, while ongoing routing task is in progress, and therefore after a timeout outdated traffic can be safely removed from memory. Finally, the creation of Quality-ofService (QoS) path, that is, path with bandwidth, delay, and connection time [SRV] requirements, requires that the path is memorized in order to optimize the traffic flow and satisfy QoS criteria. This certainly includes the use of the best path found in the search process. Once destination is reached, the optimal path can be reported back to source. f) Guaranteed message delivery. Delivery rate [BMJHJ] is the ratio of numbers of messages received by destination and sent by senders. The primary goal of every routing scheme is to delivery the message, and the best assurance one can offer is to design routing scheme that will guarantee delivery. Wireless networks normally use single frequency communication model where a message intended for a neighbor is heard by all other neighbors within transmission radius of sender. Collisions are normally occurring in medium access schemes mostly used, such as IEEE 802.11. The guaranteed delivery property assumes the application of an ideal, collision free, medium access scheme, such as time division multiple access, or acknowledgement/retransmission scheme that is assumed to be always successful otherwise. g) Scalability.

14

The routing algorithms should perform well for wireless networks with arbitrary number of nodes. Sensor and rooftop networks, for instance, have hundreds or thousands of nodes. Scalable single-path strategies, such as shortest-path, have O( n ) overhead, where n is the number of nodes in the network. While other characteristics of each algorithms are easily detected, scalability is sometimes judgmental, and/or dependent on performance evaluation outcome. We shall apply a simplified (although arguable) criterion, that a routing scheme is scalable if it is loopfree, localized, and single-path. Note that, several schemes are proved to guarantee the messages delivery (and to be loop free) in the static case. It is not clear how these schemes handle loops and perform delivery in the case of node mobility. We name these loops due to the position of some nodes as mobility-caused loops. These loops are in general temporary loops that appear because some nodes move in a position that causes the packet to loop. This situation cannot be easily detected because it arises after the direction for packet has been chosen. In this work we classify as loop free and delivery guarantee, as traditionally done, all schemes that are proved to be loop free and which guarantee the message delivery, even if they are not proved for the mobility-caused loops. h) Robustness: The use of nodes position for routing poses evident problems in terms of reliability. The accuracy of destination position is an important problem to consider. In some cases the destination is a fixed node (such as monitoring center known to all nodes, or the geographic area that is monitored), some networks are static which makes the problem straightforward, while the problem of designing location updates schemes to enable efficient routing in mobile ad hoc network appears to be more difficult than routing itself (see a recent survey [S4]) and will not be discussed here unless it is integral part of presented method. For small networks, in the absence of any useful information about destination location (that is, a clever location update scheme), the following simple strategy can be applied. If message is reasonably short, it can be broadcasted (that is, flooded), using an optimal broadcasting scheme [PL, QVL, SSZ]. If message is relatively long then destination search (or route discovery [BJMHJ]) can be initiated, which is a task of broadcasting short search message. Destination then reports back to source by routing a short message containing its position. The source then is able to route full

15

message toward accurate position of destination. However, in large networks, the algorithms that assume that the position of destination is reasonably accurate are not able to deal with eventual position deviation, and impose high mobility tracking overhead. More robust and scalable routing algorithms must, by design, be able to cope with the network dynamicity or can have backup strategies that allow reaching a node even when the node deviated from the known position. Another aspect of robust algorithms is their ability to deliver message when communication model deviates from unit graph, due to obstacles or noise. One such model is investigated in[BFNO].

16

4.Components of position based routing:

The major components of position based routing are the location services for knowing the location of the nodes and the forwarding strategies used to send messages to the different nodes.

4.1 Location Services

Position-based routing algorithms eliminate some of the limitations of topology-based routing by using additional information. The sender of a packet to determine the position of the destination and to include it in the packets destination address uses a location service. Position-based routing thus does not require the establishment or maintenance of routes. (Forwarding Strategy) In order to learn the current position of a specific node, the help of a location service is needed. Location services can be classified according to how many nodes host the service. These are of the following four categories: Some-for-some Some-for-all All-for-some All-for-all 1.Distance Routing Effect Algorithm for Mobility (DREAM) 2.Quorum-Based Location Service 3.Grid Location Service (GLS) 4.Homezone 17

Different types of Location Services are as follows: -

4.1.1. Distance Routing Effect Algorithm for Mobility (DREAM): It is an All-for-all kind of location service. Each node maintains a position database that stores position information about the other node that is part of the networks. The information consists of a node identifier, the direction of and distance to the node and a time value. Each node regularly floods packets to update the position information maintained by the other nodes. It uses temporal resolution as well as spatial resolution. DREAM proactively distributes location information. DREAM requires an all-for-all position service (each node carries a location table for each other node), and requests are forwarded in the right direction. Distance Effect: In DREAM the closer nodes are updated more frequently. The age field in location update is of primary importance and is updated frequently than the other fields. Mobility Effect: The rate of location update is controlled by mobility factor in the update table. There is no bandwidth wastage if there is no movement performed by any node. Routing policy: If there is no entry for destination change of a node in the table, then the data is flooded i.e. broadcasted. Otherwise forward the data to m neighbors in the direction of destination.

18

Figure 4: ** All for All approach.

4.1.2. Quorum Based Location Service:

It is known from information replication. The update and request from a particular node is performed on different node subsets. If subsets of different nodes intersect, then up to date information can always be found about the nodes.

Figure 5: Quorum-based location service.

Quorum based location service uses the some-for-some scheme. The node subsets host the position databases of the nodes. They act as the virtual backbone

19

between those nodes (non-position-based routing algorithm) .In this type of location service the nodes send the position update and query to the nearest backbone node. The backbone node contacts the nodes of a (usually different) quorum. The movement of backbone node is an important and has to be dealt with carefully. To solve this problem the topology of the backbone will be rearranged. If a backbone node has been disconnected from the network for more than a threshold amount of time, a new node will be chosen as the replacement of the backbone node.

4.1.3. Grid Location Service: (GLS):

The Grid Location Service (GLS) uses an ** All for some approach. It is also called as Geographic Forwarding where each node maintains its position using GPS and broadcast HELLO packet to its neighbors. HELLOW Source ID Source location Neighbor list: IDs and Location Forwarding Pointers Source speed

In Geographic Forwarding each node maintains a routing table for all nodes within two hops. It forwards a packet to the neighbor node closest to the destination node.

4.1.4. Homezone:
This kind of location service consists of a virtual homezone where position information for a node is stored. Applying a well-known hash function to the node identifier can derive the position C of the homezone for a node. All nodes within a disk

20

with radius R centered at C have to maintain position information for the node. If the homezone is sparsely populated, R may have to be increased.

Comparison of the different location services:

21

4.2 Forwarding strategies:

The Basic Principles of Forwarding Strategy are: 1.Forwarding decision: The forwarding decision mainly relies on the position of a packets destination and the position of nodes intermediate one-hop neighbors. 2.Position of the destination: The position of the destination node is contained in the header of packet to be forwarded. 3.Position update: The sending node has to keep track of the Intermediate node if it has an accurate position of the destination The different forwarding strategies are as follows. 1. Greedy packet forwarding. 2. Restricted directional flooding 3. Hierarchical routing

22

4.2.1 Greedy packet forwarding:

In a localized routing scheme, node S, currently holding the message, is aware only about the position of its neighbors within the transmission radius and destination D (indicated by black circles in Fig. ). Takagi and Kleinrock proposed the first position-based routing scheme, based on the notion of progress. Given a transmitting node S, the progress of a node A is defined as the projection onto the line connecting S and D. In the Most Forward within Radius (MFR) scheme the packet is forwarded to a neighbor whose progress is maximal (e.g., node M in Fig. ). Nelson and Kleinrock also discussed a random progress method (choosing at random one of the nodes with progress, and adjusting the transmission radius to reach that node), arguing that there is a trade-off between progress and transmission success. Hou and Li discussed the Nearest Forward Progress (NFP) method (selecting node N in Fig. 1). Finn proposed the greedy routing scheme based on geographic distance. S selects neighboring node G (Fig. ) that is closest to the destination among its neighbors. Only neighbors closer to the destination than S are

23

considered. Otherwise, there is a lack of advance, and the method fails. A variant of this method is called the Geographic Distance Routing (GEDIR) scheme. In this variant, applied on other schemes as well, all neighbors are considered, and the message is dropped if the best choice for a current node is to return the message to the node the message came from (stoppage criterion indicating lack of advance). The Nearest Closer (NC) method was proposed in (node N in Fig. ). In the compass routing method (also referred to as the DIR method) proposed by Kranakis, Singh, and Urrutia , message m is forwarded to neighbor A (Fig. ) such that direction SA is closest to direction SD (i.e., the angle ASD is minimized).

24

4.2.2 Restricted directional flooding:

DREAM - Distance Routing Effect Algorithm for Mobility

25

It is a location-based algorithm that makes use of the distance effect. This means, that two nodes appear to move slower with respect to each other with increasing distance. Thus, location information for distant nodes does not need to be updated in such an accurate and frequent way, as for close nodes. Each node has a routing table with location information about each other node. DREAM can be considered proactive, since location information must be disseminated (the method of location determination is not specified,

so a separate location service, like GPS may be required). On sending a message, a direction is determined by using the location of the destination. Then, the message is passed to all neighbour nodes in that direction. This method is more related to reactive protocols, as the route is not fixed in advance. Distance and mobility of a node determine the frequency of location updates. A fast moving node sends location control messages much more often than a slow one. Also the messages are sent with different lifetimes (in terms of hops) and short-lived messages are sent much more frequent than long-lived. The long-lived messages will reach far away nodes, but are sent much less frequent. This leads to a bandwidth and energy efficient protocol. Although the routes are not fixed in advance, there is no setup-delay. Basagni et. al. claim that this protocol is inherently loop-free, since the messages travel away from the node into a specific direction. This could be questioned, since in a network with very high mobility, the target direction can change, even back to a node who has sent the message already. Another problem is, that location table entries may be stale and that no close neighbour in the required direction can be found (e.g. due to lack of connectivity). An unspecified Recovery () routine should be called in these cases. The authors chose to use flooding in their prototype implementation. There was not much more work on DREAM, but other routing schemes such as LAR (cf. A.20) or FSR (cf. A.14) did pick up some concepts of DREAM. In this forwarding strategy the sender will forward the packet to all one-hop neighbors that lie in the direction of destination. Expected region is a circle around the 26

position of destination, as it is known to source. The radius r of the expected region is set to (t1-t0)*Vmax, where t1 is the current time, t0 is the timestamp of the position information source has about destination, and Vmax is the maximum speed that a node may travel in the ad hoc network. The line between source and destination and the angle f defines the direction toward destination.

27

B) LAR - Location Aided Routing Location Aided Routing is an enhancement to flooding algorithms to reduce flooding overhead. Most on-demand schemes, including DSR and AODV use flooding to obtain a route to the destination. This flooding results in significant overhead. LAR now aims to reduce the overhead to send the route requests only into a specific area, which is likely to contain the destination. For this purpose the notions of expected zone and request zone is introduced. The expected zone covers the area, where the destination node is expected, according to the currently known information like: -Location at some time t (this will be the center of the expected zone) - Speed at time - Direction at time t Of course, this extrapolation of the state of the node at time t, does not need to be accurate at some later time t0, but it provides a good start. Since the expected zone does not need to contain the source node, a larger area than the expected zone must be covered by the flooding, including all possible nodes on the way from the source to the expected zone. This expanded expected zone is called request zone and is used to restrict the flooding, i.e. only nodes that are part of the request zone forward a route request. On unsuccessful route discoveries, the request zone may need to be expanded further, possibly covering the whole network. Such subsequent route requests increase the initial latency for connections. This results in a tradeoff between reduced overhead and increased latency, and needs to be balanced carefully. Depending on the scheme used, a sender needs to include a specification of its request zone in its route request such that nodes receiving the request, can determine, whether they are within the zone or not. A node replying with a route will include its coordinates along with the current time (and possibly other parameters like speed and direction) in the reply, so that the sender will have its coordinates (at that time) for future requests. Suggested improvements include adoption of the request zone on the fly by the intermediate nodes of the route request. More flexible forms of request zones may be used and location information can be piggybacked to any node, to keep location information more accurate within the network.

28

Figure 6: location aided routing In the LAR strategy each node knows its own location in every moment. Routing is done using the last known location + an assumption The route discovery is initiated when: - S doesnt know a route to D. - Previous route from S to D is broken.

4.2.3. Hierarchical routing:

A) TLR/TRR/AGPF: Terminodes Routing TLR (Terminode Local Routing) This forwarding strategy uses a proactive routing scheme if the destination is close to the source node. TRR (Terminode Remote Routing) TRR allows data to be sent to non-TLR-reachable destination.

29

Terminode routing is developed at the EPFL in Switzerland. The project aim is to develop a system that is capable of wide area Ad Hoc Routing. The project did explicitly choose an independent roadmap from the IETF MANET working group efforts. Support for IP or interoperability are not very important requirements. Still, Terminode Routing will work with IP in most cases, though. Routing between terminodes is a hybrid process. First the packets are routed based on geographic position. The target address used in this routing is called LDA (location dependent address). From the target LDA the closest friend-node is computed and the packet is passed to it. A friend is a selected node in close, but not necessarily direct communication range. If the target node for the packet is among the friends of the node holding the packet, a local routing method is used to pass the packet to its destination. As position based routing needs some kind of position service, terminodes use the concept of a virtual home region (VHR), which is a some-forsome approach .For each node, there exists such a home region, which is specified by a fixed position and a radius. The region can be determined by a hash function over the node's id. Each node within the VHR of a certain node must maintain the current position of this node, so that other nodes can obtain it. The position-based routing method is called AGPF (anchored path geodesic packet forwarding). As a simple greedy forwarding mechanism doesn't work in many situations (i.e. running into a local maximum), the concepts of anchors are used. To avoid running into a maximum, the route is oriented on a set of anchors along the path. An anchor is just a specific location. The anchored path is 30

determined by the source using FAPD (friend-assisted path discovery) and included into the packet (similar to source routing). FAPD is based on small world graphs[87]. Alternatively, the path can be determined by DRD (directed random discovery), which just sends the packet to a set of neighbours whose angle is the smallest to the right direction. The local routing method is no longer based on position information, but only on a unique node identifier, the target id. A two hop neighbourhood information is maintained by each node by using HELLO packets. If the neighbourhood is known and a packet can utilize local routing (i.e. the target is known to the node which received the packet), a path discovery is initiated to direct the packet to the destination.

31

Table 2: Comparison of forwarding strategies.

Terminode routing was compared against DSR in simulations, using scenarios which were designed for the use case of terminode routing (i.e. large areas with large distances, some nodes clustered, with few roaming nodes). In these scenarios terminode routing outperformed DSR by nearly an order of magnitude.

B) Grid routing: 32

It is similar to Terminodes Routing .In this method a proactive distance vector routing is used at local level. If the destination node is too far, Intermediate Node Forwarding (INF) is used for long-distance routing. Difference between Terminodes Routing and grid routing is that it allows nodes that do not know their own position to participate in the ad hoc networks.

5. Appendix

5.1. Abbreviations:
ABR - Associativity Based Routing ADV - Adaptive Distance Vector Routing AODV - Ad Hoc On Demand Distance Vector Routing Protocol CBRP - Cluster Based Routing Protocol CGSR - Clusterhead Gateway Switch Routing CEDAR - Core-Extraction Distributed Ad Hoc Routing DDR - Distributed Dynamic Routing Algorithm DREAM - Distance Routing Effect Algorithm for Mobility DSDV - Destination Sequenced Distance Vector Routing Protocol DSR -Dynamic Source Routing DST - Distributed Spanning Tree Protocol FORP - Flow Oriented Routing Protocol

33

FSLS - Fuzzy Sighted Link State Algorithms FSR - Fisheye State Routing GEDIR - Geographic Distance Routing GPSR - Greedy Perimeter Stateless Routing GSR - Global State Routing HSR - Hierarchical State Routing . LANMAR - Landmark Routing Protocol LAR - Location Aided Routing LMR - Lightweight Mobile Routing LRR - Link Reversal Routing OLSR - Optimized Link State Routing SSA - Signal Stability-Based Adaptive Routing STAR - Source Tree Adaptive Routing TBRPF - Topology Broadcast Based on Reverse Path Forwarding TLR/TRR/AGPF - Terminode Routing TORA - Temporally Ordered Routing Algorithm WAR - Witness Aided Routing WRP - Wireless Routing Protocol ZRP - Zone Routing Protocol .

34

5.2. Definitions
We will use several phrases and terms in the following sections, which will be defined for use in our context, first. Active Receiver: A node that is receiving data at the moment or is part of a session and likely to receive data in the near future. The term active receiver is only used in ADV . Anchor: A certain geographical position in the network's area. Anchors are used for geographical routing with Terminodes. Beacon, HELLO message: A (usually) periodic local broadcast message emitted from a node, destined for its neighbours to announce itself in the neighbourhood. In some routing protocols, such a beacon may carry additional information. Bordercast:

35

A term from the Zone Routing Protocol ZRP. A message is transmitted to one or more nodes on the border of the current routing zone, where it might be transmitted further. Broadcast, local: A local broadcast is a broadcast message, which can be received from any node within reach of the sender. It is not intended to be retransmitted by the receivers.

Broadcast, limited: A limited broadcast can be retransmitted, but only to a subset of nodes in the network usually into a certain direction. Broadcast, netwide: This broadcast is retransmitted, until every node in the whole network has received the message. Cluster: A group of nodes, that act together in some way. Usually a cluster is represented by a single node, the Clusterhead. Clustering is used in hierarchical routing. Clusterhead: The representative node of a cluster. On a higher routing level, routing happens between the clusterheads. On the next higher level again clusters will be built but out of the clusterheads of the previous level. Distance Effect: The distance effect is that two nodes appear to move slower with respect to each other if they are more distant. DREAM (and certain other protocols) make use of that fact. Routing or position information for distant node does not need to be as accurate as for short distance nodes.

36

Distance Vector Routing: Simple, table based routing. Each destination is entered into the routing table with the next hop and a distance metric. The topology of the network is unknown. Expected Zone: The area in the network, where a certain node is expected to be. The expected zone is predicted from the last known movement characteristics of a node. The expected zone is used to derive a request zone, needed for location based protocols.

Flooding: A message is flooded through the whole network. This is another term for a netwide broadcast. Friend Node: In Terminode routing a node, which is in close vicinity, but does not need to be in direct communication range. A friend helps determining an anchored path in AGPF. Gateway: A node within a Cluster, often part of more than one clusters, which route messages from one cluster to another. GPS: The geographic position system, a satellite based position service operated by the US military. It enables a GPS receiver to determine its position. Group Mobility:

37

Nodes can form groups and move together as a group. This is a likely event in realistic scenarios and needs to be modeled in the scenario model. Group mobility has significant impact on routing performance, depending on the protocol and it's ability to handle group mobility. IEEE 802.11: IEEE standard family for wireless LAN communication. It defines the distributed coordinate function (DCF) or the point coordinate function (PCF) as channel allocation method. PCF can not be used easily in Ad Hoc Networks, since it would require a central instance (like an Access Point), but DCF is very common. DCF defines a RTS/CTS (request to send/clear to send) handshake to allocate a channel, thus circumventing the hidden terminal problem.

ILS: An idealized link state algorithm. Such an algorithm is mentioned in several papers as a reference for comparison, but is never specified in detail. Link Layer Notification: A mechanism, that allows the routing module to be notified of local link breaks or new links (a node moves out of reach, or a node moved into reach) from the link layer. Location Dependent Address: An address, which depends on the geographical location of a node. It must be determined by a location service. Location based routing protocols, like Terminode/AGPF make use of LDAs. Location Service: See Position Service. MANET:

38

Abbreviation for Mobile Ad Hoc Network, also name of the corresponding IETF working group.

Multipoint Relay: A dedicated node, that relays traffic for other nodes in OLSR Neighbour: Any node within direct communication range. Node: A device, capable of communication over a wireless link and attached to some (in most cases mobile) unit, like a person or a car. Nodes are members of the network.

to overhear: A node can overhear messages not destined for it, by setting its interface into promiscious mode. A node can benefit from routing information not explicitly for it. Additional routes may be learned and routes may be updated before the routes are needed and fewer route discovery processes may be needed. DSR makes use of this feature. Parent Node: The node's current uplink in a route. See also Precursor. Partial Topology: Several link state routing protocols do not maintain full topology information which would use far more ressources, but only partial topology, sufficient for efficient routing. Position Service:

39

A service, that can provide positional information for nodes in a mobile network. The position service needs to provide the position of any node to any node (GPS only provides the position for the node itself). Depending on how the position data is obtained, stored and distributed, there are all-for-all, some-for-all and some-for-some types of position services. Precursor: The precursor node in the route to a destination. If a route becomes invalid, the precursors may need to be notified to update their routing entries. AODV explicitly uses a precursor list. Proactive: Routing protocols are considered as proactive if they constantly maintain routing information for all routes, regardless, if in use or not. Maintenance can be eventdriven (also named table driven) or in regular intervals. Reactive: Routing protocols, which obtain and maintain only routes that are currently needed, are called reactive or on demand. They can cache learned routes, but if a route is unknown, a route discovery process needs to be initiated. Request Zone: A geographical zone in the network, that covers all the nodes a route request should be sent to. The request zone is used in geographical routing algorithms, which results in a limited broadcast. Rooftop Network: A static Ad Hoc network. Nodes are deployed on rooftops, but don't move, once deployed. They still need to organize themselves in an Ad Hoc fashion. We do not focus on rooftop networks.

40

Route Cache: A local cache in a node used by reactive routing protocols, to cache discovered routes. The routes will eventually time out from the cache, or be expunged if the route is detected to be invalid. Routing Loop: The route forms a loop, such that packets are routed in the loop and possibly never reach their destination if the loop persists. Routing loops need to be avoided for successful routing. Route Request: Important part of a route discovery process. A route request is usually a netwide broadcast message destined for the target node of the required route. If the destination receives the route request, it will answer with a route reply. The route request messages (also named broadcast query) are abbreviated as RREQ, REQ or QRY. Route Reply: The answer to a route request, destined for the source of the request. The route will be set up during the travel of the or can be carried directly in the route reply. This is a unicast message. It is abbreviated as RREP, RPY or REPLY. Route Error: This message type indicates a broken, stale or otherwise unusable route. It is emitted from the node, which detected the broken route and can be unicast or broadcast. It is abbreviated RERR or ERROR. Scenario: A scenario consists of a set of nodes (not necessarily of the same type or of a fixed number), that communicate and move according to the rules of the scenario. Scenarios may be very tight specified but usually are not. Scenarios are used in simulations to evaluate routing protocols.

41

Scope: A term from FSR . A scope is defined by the distance in number of hops from a node. Within each scope different update policies for routing information applies. Sensor Network: An Ad Hoc Network of tiny sensoric nodes, that are deployed in the target area. The measured data is transmitted in an Ad Hoc fashion to some collecting node. Sensor networks are rather static, but have only limited transmission ranges and even more limited power capacity.

Sequence Numbers: DSDV introduced destination sequence numbers for routes in the routing table. The sequence numbers prevent old, stale routes from being entered into the routing table. Many other routing protocols have adapted this method. Source Tree: A topology graph, representing the current routes from a source to any destination. Source Trees are used in several link state protocols, most notably STAR . Terminode: A term for the combination of a terminal and node, which is the common member of Ad Hoc Networks. It was created by the founders of the Terminode Project, a long term Ad Hoc WAN project at the EPFL in Switzerland. Virtual Home Region: The Position Service suggested for Terminode routing.

42

Virtual Link: In hierarchical routing methods, links on higher layers, than the bottom (physical) layer are called virtual links. The need to be mapped to a set of physical links on the bottom layer for the actual communication. Zone: Some area or set of nodes, that interact in a certain way. In ZRP, a zone is defined by the zone radius in terms of hops, in DDR a zone covers a tree in the forest routing topology.

5.3. Performance Metrics

Hop count: The number of hops, a packet has to travel to reach its destination. End-to-End delay: The time interval between sending a packet, and the reception of the packet at the destination. Route setup delay: The delay a packet can not be sent from a node, until the route is set up.

43

Overhead: In general overhead is the amount of data transmitted, which is no payload data. There are many different types of overhead. Overhead is usually given as a ratio of total data and payload/useful data. Routing Protocol Overhead: The data, which is sent to maintain or build routes. Retransmission Overhead: The additional data transmitted, due to retransmission of lost or garbled packets.

Suboptimal Route Overhead: This type of overhead was introduced in some papers, to reflect the overhead due to subobtimal routes (i.e. routes with a longer hop count, than necessary). Total Overhead: This type of overhead should include everything, including overhead due to suboptimal routes. Utilization: The utilization of the available network capacity. Delivery Ratio: The amount of packets actually delivered versus those being sent. The quality of a routing strategy can be well measured against the delivery ratio.

44

References
[1] Charles E. Perkins, Ad Hoc Networking, Addison-Wesley, 2001. [2] \The bluetooth special interest group," http://www.bluetooth.com/. [3] Laura Marie Feeney, \A taxonomy for routing protocols in mobile ad hoc networks," Tech. Rep., Swedish Institute of Computer Science, Box 1263, SE-164 29 Kista, Sweden http://www.sics.se/ lmfeeney/, October 1999. [4] C.-K. Toh Elizabeth M. Royer, \Review of current routing protocols for ad-hoc mobile wireless networks," Tech. Rep., UCSB, 1999. 48 45

46