Toward Strongly Connected Clustering Structure in Vehicular Ad
- Slides: 13
Toward Strongly Connected Clustering Structure in Vehicular Ad hoc Networks Zaydoun Y. Rawshdeh, Syed Masud Mahmud Electrical and Computer Engineering Department Wayne State University, Detroit, MI, USA Presented by: Sanaz Khakpour Master of Computer Science Student 1/5/2022 1
Objectives • Use clustering techniques in order to decrease the dynamic topology of VANETs as much as possible. • Cluster the nodes with most similar mobility pattern using direction, location, and speed. • Partitioning the network to minimum number of clusters. • Using a multi-metric election technique to choose the best cluster head. • Increase cluster stability considering changes in the network topology which have direct effect on stability. 1/5/2022 2
Identifying Candidate Cluster Member • Degree of speed difference is a key feature to build stable clusters. • The position information (sent in periodic messages) of vehicles is being used to build neighbourhood relationship (r-neighbour). • Nodal degree is the total number of r-neighbours of a node. • Neighbour nodes moving in the same direction are supposed to be candidate cluster members (CCM). • Neighbours are classified to SN (r-distance, same direction, close speed) and UN. • All SN which do not belong to other clusters are CCM. 1/5/2022 3
Identifying Candidate Cluster Member • 1/5/2022 4
Protocol Structure • Control channel: is being used to send periodic messages and gain information about neighbours. (Transmission range R). • Service channel: is used to create cluster and send intra-cluster messages and cluster management. (Transmission range r < R). • Because R=4 r, vehicles can obtain complete information about their neighbours (can be beyond cluster boundaries) • Any vehicle can understand if its speed is less than all its nonclustered neighbours in R distance range. That vehicle is supposed to start cluster formation. 1/5/2022 5
Cluster Radius • DSRC (Dedicated Short-Range Communications) is a multi-channel interface with various transmission ranges. • Neighbourhood definition depends on the used channels. • Vehicles u and v are neighbours in control channel’s perspective. But u and w are neighbours from the perspective of both channels. 1/5/2022 6
Cluster Formation • Each vehicle keeps a list of 2 -r neighbours at time t (Γ(t)). • Γ(t) is divided into Γ(t)_G and Γ(t)_L which are vehicles with greater and lower speeds respectively. • The vehicle with lowest speed among its neighbours starts cluster formation. It is called cluster originating vehicle (COV). • COV sends its ID to all Γ(t)_G as temporary cluster ID. All non clustered members set the cluster ID. • Vehicles calculate their suitability to be a CH and announce it if their value is higher than previously received values. Suitability value is compared with only r-neighbour members of Γ(t)_G of COV. 1/5/2022 7
Cluster Rules • Vehicles that can’t connect to the cluster stay non-clustered (default state) and start cluster formation process again. • A node joins cluster if its relative speed to CH is in the threshold. • The members should stay in r-distance range. Otherwise, they will lose their membership. • Two clusters can merge if: ü The distance between CHs are less than r. ü The difference between average speed and both CH’s speed is in a threshold. 1/5/2022 8
Cluster Head Selection • 1/5/2022 9
Cluster Head Selection • 1/5/2022 10
Simulation Results • 1/5/2022 11
Simulation Results 1/5/2022 12
Questions • What parameters are used for calculating mobility metric? • What are Γ(t)_G and Γ(t)_L in cluster formation process? • What is he paper’s most important objective? 1/5/2022 13
- Flat clustering
- Divisive hierarchical clustering example
- Euclidean distance rumus
- Number of strongly connected components
- Strongly connected graph
- Strongly connected components
- Kosaraju algorithm strongly connected components
- Strongly connected
- Strongly connected components
- Strongly connected components
- Law of conservation of momentum
- Extricación rápida
- Vehicular cloud
- In a ∆-connected source feeding a y-connected load