Τμήμα Πληροφορικής και Τηλεπικοινωνιών, ΕΚΠΑ - Προηγμένες ικτυακές Τεχνολογίες - 2016 1 Adaptive Message Routing and Information Acquisition in Dynamic Mobile Networking Environments I. Manolopoulos, K. Kontovasilis, I. Stavrakakis, S. Thomopoulos, On Demand Beaconing: Periodic and Adaptive Policies for Effective Routing in Diverse Mobile Topologies, 1Ad Hoc Networks Journal, Vol.36, Part 1, pp. 35 48, Jan. 2016. (e class) I. Manolopoulos, Kimon Kontovasilis, I. Stavrakakis, S. Thomopoulos, MAD: A Dynamically Adjustable Hybrid Location and Motion based Routing Protocol for VANETs, 7th International Symposium on Wireless Communication Systems (ISWCS'10), Sept 19 22, 2010, York, UK. http://cgi.di.uoa.gr/~istavrak/publications/2010iswcs.mkst.pdf 2 Main contributions Two parts: Adaptive Message Routing Adaptive Information Acquisition (via beaconing) Both proposed approaches utilize the innovative notion of Message Retaining Time of a node (RT) RT captures environmental conditions (density and mobility), as well as own nodal parameters (location and motion). RT shapes decisions to forward a message or not and to which node decisions to ask for (local) information or not (i.e., the frequency of on-demand beaconing schemes)
3 Mobile, Diverse and Dynamic Networking Environments Decisions should be based on: Local environment conditions Nodal characteristics Current node Most When Which is the powerful Best appropriate neighbor to Best directedtime select to as decide next located for next hop? hop? Destination Most cooperative 4 Mobile, Diverse and Dynamic Networking Environments Decisions should be based on: Local environment conditions many forwarding opportunities Nodal characteristics Nodes should decide frequently for next hop Forwarding decisions based on nodes relative position
5 Mobile, Diverse and Dynamic Networking Environments Decisions should be based on: Local environment conditions few forwarding opportunities Nodal characteristics routing loop 6 Mobile, Diverse and Dynamic Networking Environments Decisions should be based on: Local environment conditions few forwarding opportunities Nodal characteristics routing fault
7 Mobile, Diverse and Dynamic Networking Environments Decisions should be based on: Local environment conditions few forwarding opportunities Nodal characteristics Nodes should decide not very frequently for next hop Forwarding decisions based on nodes motion (e.g. direction) 8 Topologies' and Protocols' Taxonomy Limitations Highly Well-connected networks MANETs with intense mobility Infrastructure-based networks effective routing protocols should cope with diverse conditions by properly invoking (and dynamically adjusting) both the forward and carry actions. Maximum Advance Decision-MAD i protocol
9 PART A : Maximum Advance Decision (MAD) Routing Protocol Based on the Advance Metric associated with a node measures the effectiveness of the node in advancing the message towards its destination represents the rate of approaching the destination At properly defined check points the message carrying node: Collects location and motion information of the neighbor nodes (advertized by the nodes via a beacon based process). Calculates l the associated advance metrics Forwards the message to the winning neighbor (highest value of the metric), if any (A locally aware protocol) 10 Advance Metric Forward Action Progress metric: message spatial transposition towards the final destination (units of length) Progress metric due to forward action:
Advance Metric Carry Action 11 Progress metric due to carry action: Advance Metric 12 Total Progress metric: Advance metric: progress rate (units of speed) for small values of t
13 Advance Metric Properties When For small retaining times the location factor is more important than motion When For large retaining times the motion factor is more important than location The estimated retaining time acts as parameter for tuning the relative importance of the forward and carry actions 14 Next Hop Selection The relative merit of neighbor n over current node c is expressed by the difference T is a common value of retaining time for the pair c, n The value of is only due to the carry effect If, node n becomes a next hop candidate The neighbor that maximizes the advance difference is selected as the next hop node
15 The Retaining Time Estimation Employs mobility parameters collected at a decision time: Straight line motion projected for the future Mobility parameters refreshed at each check individually First approach (simple): 16 The Retaining Time Estimation Second approach (refined): Density and speed parameters are involved to all direction of motion Distinguish different directions towards the destination With
17 Simulation Setup and Parameters Open rectangular area (10km x 10km) OPNET simulator Source and Destination Static nodes Placed at diagonally opposite corners Mobile Nodes r = 250 m 0m 1000 Source 10000m Destination Random Way-Point Mobility Model (basic case const. speed, zero pausing times) Abstract PHY/MAC layer, constant time interval Td = 0.2 sec (in worst case conditions) Performance Metrics End-to-End delay - Number of hops 100 packets from the source to the destination Comparison MFR (forward-based, greedy) MoVe (carry-based) DGR (static hybrid) ~ 0.9 position weight and 0.1 motion weight MAD (dynamic hybrid) 18 Simulation Results End-to-end Delay Low density (75 nodes~0.15 neighbors) High density (4750 nodes~9.3 neighbors)
19 Simulation Results Number of Hops Low density (75 nodes~0.15 neighbors) High density (4750 nodes~9.3 neighbors) 20 PART B On-Demand Beaconing: Periodic and Adaptive Policies Beacon Mechanisms Enable the collection of neighbor-related status information, needed by locally aware protocols receiver-initiated initiated beaconing / on-demand beaconing On-demand beaconing better in highly diverse topologies Up-to-date information No need for information validating mechanisms Nodal energy saving
21 On-demand Beaconing Patterns Periodic on-demand beaconing Suitable for any routing protocol (beacon-based) How frequent should beacons be? Trade-off between quality of neighborhood perception (performance) and signaling overhead Adaptive on-demand beaconing Intelligent adaptation of the inter-beacon intervals Appropriate support from the routing protocol (parameters-metric) 22 Periodic On-Demand Beaconing How frequent should beacons be? Analyze the impact of the period on the depiction of neighborhood, h as the node sweeps an area due to its motion Select the period that achieves a balance (certain ratio) of misses (if large) and overlaps (if small)
Periodic On-Demand Beaconing Upper and Lower Bounds for Beacon Period 23 Area with sensed nodes: Area with non-sensed nodes: Common overlap area: (a): where (b, c): Periodic On-Demand Beaconing Upper and Lower Bounds for Beacon Period 24 Loss factor: Upper bound: or (fraction of nodes sensed multiple times) Lower bound: or
Adaptive On-Demand Beaconing 25 When to beacon? Need to take into account: The appropriateness of the current node Good carrier means infrequent (rare) neighborhood check The forwarding opportunities in the environment Many forwarding opportunities mean frequent neighborhood check The retaining time captures: The impact of a node s location and direction of motion The impact of nodal density and speed (mobility) Proposed rule of thumb: Inter-beacon interval = c RT (c<1) Adaptive On-Demand Beaconing 26 Conclusions 0.2 RT and 0.25 RT (original, improved) best after lots of simulations Effectiveness of the bounds on α: values of α between the bounds yield effective balance between low end-to-end delay and low number of hops and beacons. comparing adaptive vs periodic beaconing: adaptive scheme provides a better trade-off between performance and beacon messages than the periodic scheme.
Simulation Results 27 Sparse (500 nodes~1neighbor) and low mobility (10 km/h) Sparse (500 nodes~1neighbor) and high mobility (50 km/h) Dense (2550 nodes~5neighbor) and high mobility (50 km/h)