Impromptu ideas in respect of v2 v and other
Download as PPTX, PDF1 like511 views
The document proposes using a distributed and cooperative communication approach for controlling future robots. A distributed approach would provide robustness over a centralized system by avoiding single point of failures. However, the environment is unpredictable, so cooperative mechanisms are needed like storing commands that can be used when needed. The document also suggests combining centralized and distributed systems by allowing the centralized system to change dynamically. This would decrease the probability of failure enormously. It then discusses using concepts from network science like scale-free networks to help define fast routing protocols for vehicle communication networks where nodes change positions quickly.
![Introduction through VANETS
• A vehicular Ad-Hoc Network (VANET) is a technology that
uses moving vehicles as nodes in a network to create a
mobile network.
• VANET turns every participating vehicle into a wireless
router or node, and in turn create a network with a wide
range.
Goal: Provide Road Safety
Challenges in V2V:
1.Relative speed
2.Anonymity
3.Changes in position
Case Study: As More vehicles and range increase in VANET
lead to reduce access prob. by 35%.[1]](https://image.slidesharecdn.com/impromptuideasinrespectofv2vandother-151016052708-lva1-app6892/85/Impromptu-ideas-in-respect-of-v2-v-and-other-15-320.jpg)
Impromptu ideas in respect of v2 v and other
- 1. Impromptu Idea in Respect of Robotics Communication By Harshit Srivastava
- 2. A Historic Idea • In star wars “The phantom of menace” we learned one thing, • The bad robots had centralized failure, and whole army was useless when main controlling computer was destroyed.
- 3. Solution • So, for controlling of future robot a distributed communication approach should be followed. • The distributed approach requires a complex management system, but gives robustness. • But, the environment is unpredictable and can be unplanned so, a cooperative mechanism is needed. • One feature of store command and use it when needed( Can be used for high latency) • Prioritization. • The next key thing is “Prediction.”
- 4. Note: • Still, I believe the centralized and distributed communication if can be combined where the centralized system can be changed dynamically. • Then, probability of failure will decrease enormously.
- 5. Lets say! RSU== Virtual Cloud RSU== Virtual Cloud RSU== Virtual Cloud RSU== Virtual Cloud Broadcast *Safety message are priority, so they should be broadcasted
- 6. Lets say!(A more predictable) RSU== Virtual Cloud RSU== Virtual Cloud RSU== Virtual Cloud RSU== Virtual Cloud Broadcast *Safety message are priority, so they should be broadcasted
- 7. • So, in previous slide an approach towards creating a temporary centralized network through distributed system. • Now, the problems are, • How to determine super node(centralized node) • If previous super node has failed how to create new or find new super node.
- 8. Now we again look into the model RSU== Virtual Cloud RSU== Virtual Cloud RSU== Virtual Cloud RSU== Virtual Cloud Broadcast *Safety message are priority, so they should be broadcasted Greedy Node Greedy Node Greedy Node Greedy Node
- 9. • So, we can go through network science and say that if the greedy node is present in a network the graph will grow as scale free network. • This scale free network has a can be said small-world network. This network where the average shortest path length between its nodes is very small compared to the total number of nodes in the network. • Average shortest path length which grows proportionally to the logarithm of the number of nodes of the network. 𝐿(𝑁) ∝ log(𝑁)
- 10. Prediction Approach • So, for prediction approach two methods I think we can look into • Through Jaccard Coefficient where the set of similarities is measured to predict node. • Another, method is through Hidden Markov Model(Markov process and Markov chain). • Since, for HMM state at discrete moment of timein basically unknown, only physical event is observed, where the observation is a probabilistic function of hidden state and can be characterized as, • N, number of states in the model. The individual states can be denoted as 𝑠 = (𝑠1, 𝑠2 , 𝑠3 … . . , 𝑠 𝑁 ) • M, the number of distinct observation symbols. The individual symbols can be denoted as O= {𝑜1, 𝑜2 , 𝑜3 … , 𝑜 𝑀 } • The state transition probability matrix 𝐴 𝑁𝑋𝑀 = {𝑎𝑖𝑗}(1 ≤ 𝑗 ≤ 𝑁) • The observation symbol probability matrix 𝐵 𝑁𝑋𝑀 = 𝑏𝑗𝑘 𝑤ℎ𝑒𝑟𝑒 𝑏𝑗𝑘 = 𝑃 𝑂 𝑘 𝑎𝑡 𝑡 𝑆𝑡 = 𝑠𝑗 , 𝑡 = 1,2. . • The initial state distribution vector 𝜋 = {𝜋𝑖} where: 𝜋𝑖 = 𝑃 𝑆1 𝑠𝑖 , 1 ≤ 𝑖 ≤ 𝑁 • Where 𝑆𝑡 denotes the state space at 𝑡 time.
- 11. References • Shahzad A.Malik,Madad Ali Shah, Shahid A.Khan,M.Jahanzeb, Umar Farooq and Adnan, Performance Evaluation of IEEE 802.11p MAC Protocol for VANETs; Khan. Australian Journal of Basic and Applied Sciences, 4(8): 4089-4098, 2010. • Mir, Zeeshan Hameed, and Fethi Filali. "LTE and IEEE 802.11 p for vehicular networking: a performance evaluation." EURASIP Journal on Wireless Communications and Networking 2014.1 (2014): 1-15. • Robert Gallager, “Stochastic Process and its application” • Zhigang Wang, Lichuan LIU, “Protocols and applications of Ad-hoc robot wireless communication networks: Overview”, International Journal of Intelligent Control and Systems.
- 12. Thank you
- 13. • Let’ say we have any one of finite or countable infinite states(𝑠1, 𝑠2, 𝑠3 … 𝑠 𝑛) and 𝜏 denote set of states and is a state space of system for time=1,2,..T • 𝑆𝑡 denotes the state space at 𝑡 time. Then first order Markov chain prob. of the system 𝑆𝑡 at time 𝑡 and depends only on the previous state of 𝑡 − 1 as, • Type equation here.
- 15. Introduction through VANETS • A vehicular Ad-Hoc Network (VANET) is a technology that uses moving vehicles as nodes in a network to create a mobile network. • VANET turns every participating vehicle into a wireless router or node, and in turn create a network with a wide range. Goal: Provide Road Safety Challenges in V2V: 1.Relative speed 2.Anonymity 3.Changes in position Case Study: As More vehicles and range increase in VANET lead to reduce access prob. by 35%.[1]
- 16. Ideas that were floated Future Research Scope Vehicula r Cloud Mac Layer Protocol Mobility Model Image Processing Fault ToleranceVehicular Cloud Mac Layer Protocol Mobility Model Routing Algorithm Design
- 17. So what we can choose? • We can easily analyse that information hopping should be fast, so if node is lost then next super node should be found for information hopping, and that node should depend on prediction. • So, what we can do, for prediction in network, Jaccard Coefficient for predicting next super node.(jaccard algo can also be sued to refine data and predicting the best node) 𝑠𝑐𝑜𝑟𝑒 𝑥, 𝑦 = 𝑠𝑒𝑡 𝑥 ∩ 𝑠𝑒𝑡 𝑦 𝑠𝑒𝑡 𝑥 ∪ 𝑠𝑒𝑡 𝑦 Or , 𝑠𝑐𝑜𝑟𝑒 𝑥, 𝑦 = 𝑙=1 ∞ 𝛽 𝑙 |𝑝𝑎𝑡ℎ𝑠 𝑥,𝑦 𝑙 | • In which similarity of nodes are found. • If we look this network in new aspect, we should try to model it as a scale free network. • Now, if there is a network and nodes, lets take a look in respect of network science. • In this the connectivity can easily be defined according to scale free property.
- 18. • So, there should be a feature that can store protocol commands and can restore when needed.
- 19. Is scale free relevant? • Since we know, A small-world network is a network where the average shortest path length between its nodes is very small compared to the total number of nodes in the network. • Average shortest path length which grows proportionally to the logarithm of the number of nodes of the network. 𝐿(𝑁) ∝ log(𝑁) Scale free has a special property, in which structure of network remains same regardless of scale of observation.
- 20. A(v,a) B(v’,a) C(v’’,a) Now after t time B changes acceleration
- 21. • Since last slide shows the network connection of different vehicles. • To comprehend that, we have data grabber nodes(cloud), super node. • The problem here lies, if super node is dropped or changed, which one to choose to hop information. • Many, algorithms can be tried like short path and link state, but all requires destination node and they determine first path before sending the data. • So, what we can choose?
- 22. • We know scale free have fault tolerant behaviour, like if failure occurs in random and vast majority of nodes have small degree, then most probably hub will be safe. • Another property of clustering coefficient which decreases as node degree increases
- 23. What can this give? • Can help us to define fast routing protocols in which node changes quickly. • This allows a vehicle to determine whether it is the vehicle with the higher number of neighbours in the network. • Then this node vehicle can be utilized for information hoping.
- 24. Future Delivery System. • Last year Amazon has tested their drones to deliver parcels. • But, still, drone delivery system is at preliminary stage, as still I believe proper ground is not prepared for there control. • For, me the main problem is like how to control these drones when to deliver good for larger distance. • So, I think this vehicular communication can indirectly solve this in respect of virtual cloud control, in delivering command with low latency.
- 25. One Future communication • Brain to machine communication • Brain machine interface can be said as neural interface with computer. • Brain machine communication can be used as command , control and communicate with the integrated systems
- 26. Lte for Vehicular Communication
- 27. Issues • Coverage and mobility: This will rely only on nodes organised by network infrastructure. • Centralized architecture: It can not natively support V2V, but require passing through infrastructure nodes that should intercept and distribute. • Channel and transport modes have problem.
Editor's Notes
- #11: Given present state past history does not affect conditional prob. of events defined in the future and markov chain are discrete parameter of markov process whse state space is finite or infinite.