Download presentation
Presentation is loading. Please wait.
Published byBartholomew Hancock Modified over 6 years ago
1
Surviving Holes and Barriers in Geographic Data Reporting for
Wireless Sensor Networks Yangfan Zhou1, Michael R. Lyu1, and Jiangchuan Liu2 1Dept. of Comp. Sci. & Eng., The Chinese University of Hong Kong, Hong Kong 2School of Comp. Sci., Simon Fraser University, Canada
2
Outlines This is the outline of this presentation. First we will give a brief introduction on the background and the motivations of this work. Then we illustrate our proposed waypoint-based geographic data reporting protocol. And then we will demonstrate how GDRP works in an example network. Finally, we will present our simulation results and conclude this work.
3
Background and Motivations
Let’s first introduce geographic forwarding. A node will forward packet to a neighbor closer to the sink until the packets arrive. [X] But if no such neighbor can be found. That means there is a network hole or barrier ahead. [X] Traditional geographic routing will enter a detour mode where packets are forwarded along the face of the planar graph of the network.
4
Background and Motivations
The detour-mode forwarding tends to forward data packets along the boundaries of holes. Let’s see an example. The source will forward packets to u1 because u1 is closest to the sink…and so on, until the packet reaches u3. [X] u3 cannot find any node closer to the sink than itself. That means there is a network hole ahead. Packet will enter the detour mode and is routed along the face of the planar graph [X] until it reaches u8, where u8 can find a node closer to the sink. Greedy forwarding resumes. [X] We can see that the detour mode forwards packets along the boundary of a hole. [X] The path is much longer than the optimum will indicates more energy consumption…We should avoid detour mode. The path from the source to the destination is much longer than the optimum. More energy consumption in data collection.
5
Background and Motivations
■ Minimize the unnecessary detours so that the path can bypass holes and barriers. So people propose waypoint-based geographic forwarding approach, in which we select a set of waypoint and let packet go through these waypoints. Packet are transported between two adjacent waypoints with a geographic forwarding scheme. [X] The purpose is to by pass holes and barriers to minimize unnecessary detours We can see that waypoint calculation is a major task. Which is generally calculated with a trial and error approach ■ Waypoints: Calculated with a trial-and-error approach - Better and better waypoint sequences will be worked out gradually
6
Background and Motivations
Existing waypoint-based schemes, however, may result in suboptimal paths. First, existing schemes believe that when greedy forwarding is feasible between any adjacent waypoint pairs, then the path is good path. Actually, this is not enough. A greedy forwarding path may be a suboptimal path too. The left-hand-side path is a greedy forwarding path, but it is not an good path, we can see that the right-hand side path is better. [X] Second, existing schemes propose to route packets along the convex hull of the holes, this is not always a good idea, we can see that the path penetrate in between the holes are better. So we are motivated to find a better waypoint-based geographic forwarding scheme. Existing schemes may result in suboptimal paths
8
Geographic Data Reporting Protocol (GDRP)
Like most waypoint-based geographic routing schemes, our protocol, named GDRP, adopts a trial-and-error approach. [X] First, only the source and the destination are the waypoints. [X] And then after a path is found, we will check whether the path is acceptable. [X] if yes, we are done [X] Otherwise, we calculate another set of waypoint for the next round. [X] So there are three key parts in the protocol. The first is how to calculate a set of waypoint [X] Second, how to adapt the geographic routing approach in between adjacent waypoints [X] Finally, when we find a path, we should determine whether this path is an acceptable path and ends the procedure [X] We will illustrate our designs of these three parts in what follows.
9
When a Path is Acceptable
Now let’s discuss when a path is acceptable. Let’s define the so-called strongly perfect sequence first. A strongly perfect sequence is a greedy-forwarding path from the first node to the last node. Moreover, [X] all the nodes in the sequence must be in a rectangular area with length and width being l and alpha r, where l is the distance between the first node to the last node, and r is the communication range, and alpha is a constant. [X] With this definition, we say a path is an acceptable path if the path segments between any two adjacent waypoints are all strongly perfect sequences. Why?
10
When a Path is Acceptable
Energy consumption in data forwarding Let’s first define a term we will use frequently later. The topological length of a path: which is exactly the hop counts of the path. [X] This indicates the energy consumption of the data forwarding task from the source to the sink
11
When a Path is Acceptable
Because we have this lemma. The topological length of a path is linearly related to the Euclidean distance between the source and the destination if the path is a strongly perfect sequence [X] And then a corollary is that if a path is strongly perfect sequence, the topological length of the path is at most a constant times of the shortest path. The linear relationship indicates the performance guarantee.
12
When a Path is Acceptable
Maximum # of such circles Topologic length of the path
13
When a Path is Acceptable
1. The topologic length of a strongly perfect sequence 2. The topologic length of the shortest path is lower-bounded by Hence, the topologic length of a strongly perfect sequence will not be worse than times the shortest path
14
Waypoint Calculation
15
Waypoint Calculation Now if a path is not yet acceptable, we need to calculate a new set of waypoints. First, when a path is not acceptable, at least one path segment between two adjacent waypoints is not a strongly perfect sequence. Hence we can model the impact of the holes or barriers as how they make the path segment imperfect.
16
Waypoint Calculation perfect sequences: [w, u1, u2,w’] [u4, u5, u6,w’]
Let’s define perfect sequence first. Perfect sequence is similar to strongly perfect sequence. The only difference is that the last second node do not need to connect to the last node. [X] Like this, w u1 u2, w’ is a perfect sequence, where u2 is not connected with w’ [X] Similarly, this is a perfect sequence [X] And this is a perfect sequence too, which is also a strongly perfect sequence since the last two nodes are connected [X] Except the perfect sequences, we consider the other parts are those that make the whole path fail to be a strongly perfect sequence. [X] We call them detour parts, and consider them as the current knowledge of the holes and barriers. We calculate waypoint based on these detour parts. [u2, u3, u4] and [u6, u7, u8] make the whole path segment fail to be a strongly perfect sequence
17
Waypoint Calculation Problem 1: Who should be the potential waypoints
1. The last node in a detour part should be a potential waypoint 2. The first node in a detour part should be a potential waypoint Reason: The hole or barrier does not influence the path any longer from the node on Reason: The node can avoid the detour part by forwarding packets to another direction First of all, let’s see who should be the candidates. [X] Let’s again see this example. [X] We consider the last node in a detour part is a candidate. [X] This is because we consider from this node on, the hole or barrier modeled by the detour part does not have an impact on the path any more. [X] Also, we consider the first node is a detour part is a candidate. [X] The reason is that we will let the node try to bypass the hole by routing packets to another direction [X] Hence, in this example, we have four candidates
18
Waypoint Calculation Now with the candidates, we expect in the next round, we can find an acceptable path. In other words, the path segment between two adjacent waypoints should be strongly-perfect sequence. [X] So let’s construct a graph G(V,E) where V is the set of the waypoint candidates and two waypoint share an edge if the line segment between them does not intersect a known detour part. [X] Hence, the waypoint sequence for next round should be a path in this graph [X] Straightforwardly, the shortest path is the best. Hence we will find the shortest path, [X] Hence, the waypoint sequence for the next round is [w, u2, u8, w’] Hence, the waypoint sequence for the next round is [w, u2, u8, w’]
19
Geographic Routing between Waypoints
Now we talk about the third component of our protocol, how to route packet between adjacent waypoints. This is almost the same as the traditional geographic routing approach. The only difference is that we let a starting node of a detour part can change its routing direction to explore another direction of the network. Still a light-weighted protocol for sensor nodes
21
GDRP Demo This path segment make the whole path fail to be a strongly perfect sequence Nodes 1 and 2 are then potential waypoints, because they are the first node and the last node of the detour part Construct the waypoint graph Find the shortest path from the source to the sink. Hence, the new waypoint for the next round is [source, destination]
22
GDRP Demo Again, we will reach this node, since it is the starting node of a detour part, it will route the packet to another direction This path segment make the whole path fail to be a strongly perfect sequence Node 3 is then a potential waypoint. Because node 1 is the first node of two detour parts, it means it cannot bypass a hole or barrier by routing packets to another direction. So it is removed from the potential waypoint set. Construct the waypoint graph Find the shortest path from the source to the sink. Hence, the new waypoint for the next round is [source, node 3, destination]
23
GDRP Demo Now packets will be sent to node 3 first.
This path segment make the whole path fail to be a strongly perfect sequence Nodes 4 is then a potential waypoint. Construct the waypoint graph, find the shortest path from the source to the sink. Hence, the new waypoint for the next round is [source, node 2, destination]
24
GDRP Demo Now packets will be sent to node 2 first.
These path segments between adjacent waypoints are both strongly perfect sequences We are done! We can see that the resulting path is comparable to the shortest path
26
Simulation Studies Now we present the simulation part.
We compare our GDRP with GPSR, a famous geographic routing algorithm and convex-w, the most recent waypoint-based geographic forwarding protocol. In convex-w, packets are forwarded along the convex hull of the holes found.
27
Simulation Studies
28
Simulation Studies First of all, our protocol performs the best in terms of topological length of the path found we can see that when the hole and barrier number increases, the topological length of our protocol does not increase dramatically as the other two algorithms. This shows the waypoints we found are effective in bypassing holes and barriers.
29
Simulation Studies We change the node number and the communcation range of each node The results in these two figures again confirm the advantage of our protocol comparing with the other two algorithm
30
Conclusions We conclude our contribution in this part, we specifically tailor GDRP, a waypoint-based geographic forwarding protocol for energy-constraint WSNs, which can find a performance guarantee path
31
Q & A Thank you!!!
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.