Presentation is loading. Please wait.

Presentation is loading. Please wait.

Phero-Trail: Bio-inspired Routing in Underwater Sensor Networks Luiz F. Vieira, Uichin Lee, Mario Gerla UCLA.

Similar presentations


Presentation on theme: "Phero-Trail: Bio-inspired Routing in Underwater Sensor Networks Luiz F. Vieira, Uichin Lee, Mario Gerla UCLA."— Presentation transcript:

1 Phero-Trail: Bio-inspired Routing in Underwater Sensor Networks Luiz F. Vieira, Uichin Lee, Mario Gerla UCLA

2 Application Scenario Protecting critical installation such as harbor, underwater mining facility, and oil rigs.  Mobile floating sensor nodes (they move with currents)  Autonomous Underwater Vehicles (AUV) or Submarines  AUV periodically “probes” sensors to check coverage and request “refills” of holes

3 SEA Swarm Architecture Sensor Equipped Aquatic (SEA) swarm of mobile sensors:  Span a Cube (say 1x1x1 km) under the ocean surface  Enable 4D (space and time) monitoring  Dynamic monitoring in face of ocean currents Sensor nodes notify events to corresponding submarines Z X Y Event Type A Event Type B Event Type A Event Type B

4 The Problem Mobile sensors report events to submarines  Must discover/establish route to submarine Proactive (OLSR), Reactive Routing (AODV), or Sensor data collection (Directed Diffusion)  All require route discovery (flooding) and/or maintenance  Not suitable for bandwidth constrained underwater mobile sensor networks (collision + energy consumption) Geographical routing preferable, but expensive:  U/W coordinate maintenance (no GPS)  geo-location service to know the destination’s location Goal: design an efficient routing protocol for a SEA swarm

5 Solution: Phero-Trail AUV uploads updates (pheromones) to sensors along its projected trajectory on the “upper hull” of 1x1x1 cube  Periodic updates create a pheromone trail  We assume the trail length is = O(M), where M: number of hops to travel width of network

6 Phero-Trail Routing A mobile node routes the query packet vertically upwards to the convex hull Node on convex hull performs a search to find a pheromone trail. Once it finds the trail, the query travels to the end node (most recent update) The packet is sent downward to the UAV (following the “bread crumbs” on the vertical path)

7 O/H Analysis Update  Length of a pheromone trail is assumed O(M)  Up to M nodes remain connected in spite of current  Update O/H (pkt tx/sec) = MS/R = O(M)  S = UAV speed and R = Tx range (uniform sensor distrib) Routing  Vertical routing O(M) - easy due to pressure gradient  Hull plane search for trail: random walk with step size =R Prob of success = 1/M (ie M pheromones among M 2 sensors) Avg # of steps to find trail = M Travel until the end of trail = M Travel downwards to destination UAV = M  Total Routing O/H = O(M) pkt transmission

8 Related Work – Naïve Flooding Node periodically floods its current position to the entire network (M 3 overhead) M: number of hops to travel width of network Z X Y

9 Simulation Results 1 Km x 1 Km x 1 Km Submarine 5 m/s Vary network size Compared with flooding benchmark O/H = Number of transmissions to deliver a packet to UAV.

10 Simulation Results Number of pkt transmissions as a function of number of submarines (UAVs)

11 Related work - The ants Ants can explore vast areas without global view of the ground. Can find the food and bring it back to the nest. Will converge to the shortest path.

12 How can they manage such great achievement ? By leaving pheromones behind them.  Marking the area as explored  Communicating to the other ants. Aging: pheromones evaporate Double Bridge experiment Food

13 AntHocNet : Ants-inspired routing Ants will start from A the nest and look for D the food. At every step, they will upgrade the routing tables and as soon as the first one reaches the food. The best path will be known, thus allowing communication from D to A. E D B A F C Nest Food

14 AntHocNet (1/3) (1) When a data session is started at node s with destination d, s checks whether it has up-to-date routing information for d. (2) If not, it reactively sends out ant-like agents, called reactive forward ants, to look for paths to d.

15 AntHocNet (2/3) (3) These ants gather information about the quality of the path they followed, and at their arrival in d they become backward ants which trace back the path and update routing tables. (4) Once paths are set up and the data session is running, s starts to send proactive forward ants to d:

16 AntHocNet (3/3) Packets are broadcast to alternate paths with some probability, so that new paths are explored. (5) In case of link failures, nodes either try to locally repair paths, or send a warning to their neighbors such that these can update their routing tables. VERY SIMILAR TO AODV

17 Related work: P2P Based Routing FRESH (like Encounter w/out GPS coordinates):  Target Node broadcasts its ID to all those it encounters.  Source “flood searches” for contacts, ie neighbors with target ID  Source forwards packet to most recent (FRESH) contact  Differs from Phero-Trail FRESH relies on mobile “mules”; works best if destination is static Phero Trail relies on static sensors and “moving” destination (AUV) BreadCrumb:  Assume static wireless sensor networks (OK)  The originator uses Bread Crumbs to retrace the path  Very similar to the down trace of the vertical path to get to the AUV

18 Related Work –Location service Quorum Based LS Each location update is sent to a subset of nodes (update quorum, eg vertical line) Location query is sent to a subset of nodes (query quorum, eg horizontal line) The query will be resolved when their intersection is non- empty Still need geo coordinates to route to destination XYLS

19 Related Work – Hierarchical LS Location servers are chosen via a set of hash functions Area recursively divided into a hierarchy of smaller grids. For each node, one or more nodes in each grid at each level of the hierarchy are chosen as its location servers.

20 Hierarchical - Example Node updating location Server at level 2 Server at level 3 Node requesting location

21 Protocol Analysis M: number of hops to travel a width of network L; i.e., M=L / R, where R =com. range Quorum-based must store information in a 2D plane; i.e., O(M 2 ) UpdateQuery Naïve flooding O(M 3 )O(1) Quorum- based O(M 2 )

22 Protocol Analysis Hierarchical  Must first find a reference point for geographic hashing and propagate this information to every node.  Overhead of “periodical” reference point updates dominates the update/query overhead.

23 Reference Point Updates in Hierarchical Schemes Periodic reference update O/H: O(M 3 )

24 Location Service in 2D? Store location information in 2D; search and update in 2D But at the cost of vertical routing O(M) to given a location service plane Where to put a 2D plane?  Ex: Upper hull (easy to detect) Simply check local max

25 Location Service in 2D: Analysis Naïve flooding  update and query costs are O(M 2 ) Quorum-based  store information in a 1D line  Update and query costs scale as O(M) Hierarchical  Reference update, location update and query operations take O(M 2 ), O(H), and O(M) respectively.  Reference point update is still expensive!!!

26 Conclusion Presented a novel bio-inspired Underwater Routing scheme (Phero Trail)  efficient routing for a SEA swarm  Optimal performance O (M)  Outperforms naïve approaches  Can perform as well as position aware schemes (eg Hydrocast) Yet, it does not require coordinates

27 Future Work Phero Trail  Study performance in presence of currents  Study effect of multiple submarines  Vary number of sensors/sinks, the speed of sensors/sinks, the deployment area size (including various depths).  Impact of coordinates - can do better with them? Use Phero Trail as Location Server (PTLS)  Compare performance of PTLS with High-Grade, XYLS


Download ppt "Phero-Trail: Bio-inspired Routing in Underwater Sensor Networks Luiz F. Vieira, Uichin Lee, Mario Gerla UCLA."

Similar presentations


Ads by Google