Spreading Process in Multilayer Network
Nandini, Dr. Janak Kumar B. Patel, Tejender Rawat
M.Tech Student, Dept. of ECE, ASET Amity University, Haryana, India
Professor, Dept. of ECE, ASET Amity University, Haryana, India
Assistant Professor, Dept. of ECE, ASET Amity University, Haryana, India
ABSTRACT: Many systems can be designed as sets of interconnected networks or networks with multiple types of connections, they are called multilayer networks. Spreading processes such as an information propagation among users of online social networks or we can say that the diffusion of pathogens among individuals through their contact network are fundamental phenomena occurring in these networks. Information diffusion in single networks has received considerable attention from different disciplines for over a decade and spreading processes in multilayer networks is still a young research area presenting various challenging research issues. In this paper, we review the models, results and applications of multilayer spreading processes and discuss some upcoming research directions.
KEYWORDS: Multilayer network, multiplex, interconnected, spreading processes, information diffusion. I. INTRODUCTION
During designing of this basic model several attributes can be added to nodes and edges. A temporal dimension is considered and make a distinction between node on layer 2 at time ‘t0’ and the same node at time ‘t1’ and then we can add edges among these extended nodes. Connections between nodes on the different or same layers represent channels through which different types of items can propagate and giving rise to spreading processes. Basically, spreading process can refer to the diffusion of pathogens, rumours, behaviours, or the coverage of a news headline in different newsgroups and weblogs. For example, in the case of spreading of some behaviour in a community, people usually choose which behaviour to adopt. On the other side, in the case of epidemics model there is no decision made by the individuals who are infected. In the current section, we present the key concepts that may arise in the analysis of spreading processes.
Figure-1.1 Spreading processes in multilayer network [13]
‘lv’.Seed the first node of the tuple with the minimum timestamp is called initially. While this is the minimum amount
of information needed to meaningfully describe a spreading process, specific models reviewed in augment these tuples with additional parameters (a state space and a set of rules for state-transition) providing more details about the cascade.
This can be resulted by presenting the main dependent variables used in different spreading studies. The so-called epidemic threshold is one of the key observations in epidemic-like models and indicates a value of transmissibility above which the diffusion involves the whole network. Example, diffusion network is a giant component of the underlying network. It is known that in amonoplex networks the value of the epidemic threshold is closely related to the largest eigenvalue of the network’s adjacency matrix. Furthermore, recent work suggests that the epidemic threshold in a multiplex network cannot be larger than the epidemic thresholds of individual layers. In the context of interacting spreading processes in multilayer networks, two types of thresholds have recently been introduced, called survival threshold and absolute-dominance threshold: they measure if a spreading process will survive and whether it can completely remove another competing process. Another dependent variable is the infection size, generally defined as the number or fraction of nodes in the diffusion network, i.e., those reached by the spreading process. Similar terms such as outbreak size or cascade size have also been used in the literature to refer to this quantity. The Infection rate, representing the average rate of being in contact over a link, is also a frequently studied dependent variable.
There are four possibilities for an item to traverse a multilayer network.
1. Same-node Inter-layer: When a cascade switches layer but remains on the same node. Example: When a facebook post is shared on twitter by the author of the same post.
2. Other-node Inter-layer: When a cascade continues spreading to another node in another layer. Example: Exchanging mails between users with different mail accounts (like Gmail and yahoo).
3. Other-node Intra-layer: The cascade continues spreading through the same layer. Example: retweeting a post in Twitter.
4. Same-node Intra-layer: This layer is generally not considered meaningful approach and therefore omitted in all the spreading studies.
II.DESIGN AND IMPLEMENTATION
3.1 OpenStack Networking
OpenStack Networking is a separate service that can be connected individually of other OpenStack services. OpenStack services that come under this category consist of Compute, Identity, Block Storage,Image and Dashboard. OpenStack Networking services can be split amongst multiple hosts to provide flexibility and redundancy, or also configured to operate on a single node. OpenStack Networking uses a service called neutron-server for description of an Application Programmable Interface (API) to users and to pass requests to the configured network plugins for further processing. Users are able to describe network connectivity in the cloud and cloud operators are allowed to influence different networking technologies to improve and power the cloud. Networking requires access to a database to determine storage of the network configuration.
3.2 Features of OpenStack Networking
OpenStack Networking includes many technologies one would find in the data centre, comprising switching, load balancing, routing, firewalling, and virtual private networks. These features can be configured to influence open source or commercial software and offer a cloud operator with all of the tools necessary to build a functional and independent cloud. OpenStack Networking also provides a framework for third-party dealers to build on and improve the competences of the cloud.
3.2.1 Switching
bridging and OpenvSwitch. OVS is an open source virtual switch that maintains standard management interfaces and protocols containing NetFlow, SPAN, RSPAN, LACP, and 802.1q. However many of these structures are not visible to the user through the OpenStack API. In addition to VLAN tagging users can figure overlay networks in software using Layer 2 in Layer 3 tunneling protocols such as GRE or VXLAN. Open vSwitch can be used to enable communication between instances and the devices outside the control of OpenStack which contains storage devices, hardware switches, network firewalls, dedicated servers and many more. The use of Linux bridges and Open vSwitch as changing platforms for OpenStack can be establish in constructing a Virtual Switching Infrastructure.
3.2.2 Routing
OpenStack Networking delivers routing and NAT capabilities from end to end for the use of IP tables, IP forwarding and network namespaces. A network namespace is similar to chroot for the network stack. Inside a network namespace I found sockets, bound ports and interfaces that were generated in the namespace. Each network namespace has its own routing table and IP tables route that deliver filtering and Network Address Translation also identified as NAT. Network namespaces are similar to routing instances in Juniper JunOS, VRFs in Cisco or route domains in F5 BIG-IP. There is no concern of overlapping subnets between networks generated by tenants. Constructing a router within Neutron allows instances to interact and communicate with outside networks.
3.3 Preparing the Physical Infrastructure
The purpose of the cloud itself along with security requirements and available hardware will play a big part in determining the architecture of the network and OpenStack's role in the network.
The OpenStack portal www.openstack.org provides reference architectures for Neutron-based clouds that involve a combination of the following nodes:
1. Controller node 2. Network node 3. Compute node(s)
Prior to the installation of OpenStack, the physical network infrastructure must be configured to support the networks needed for an operational cloud. In the following diagram, I have highlighted the area of responsibility for the network administrator:
Figure-2.1 The area of responsibility for the network administrator [22]
3.4 Types of Network Traffic
Types of network traffic do not require dedicated interfaces and are often collapsed onto single interfaces. Depending on the chosen deployment model, the cloud architecture may spread networking services across multiple nodes. The security requirements of the enterprise deploying the cloud will often dictate how the cloud is built.
3.4.1 Management network
The management network is used for internal communication between hosts for services, such as the messaging service and database service. All hosts will communicate with each other over this network. The management network can be configured as an isolated network on a dedicated interface or combined with another network as described below.
3.4.2 API network
The API network is used to represent openstack APIs to the users of the cloud and services inside the cloud. Endpoint addresses for facilities such as Neutron, Keystone, Glance and Horizon are obtained from the API network. It is common exercise to configure a single IP address on a devoted interface that will function as the audience member address for the various services as well as the management address for the host itself.
3.4.3 External network
An external network provides Neutron routers with network access. Once a router has been configured, this network becomes the source of floating IP addresses for instances and load balancer VIPs. IP addresses in this network should be reachable by any client on the Internet.
3.4.4 Guest network
The guest network is a network dedicated to instance traffic. Options for guest networks contains local networks restricted to a particular node, flat or VLAN tagged networks or the use of virtual overlay networks made possible with GRE or VXLAN encapsulation. The interfaces used for the external and guest networks can be devoted interfaces or ones that are shared with other types of traffic each approach has its benefits and caveats.
3.5 Physical Server Connections
The number of interfaces needed per host is in need of the type of cloud being built and the security and performance requirements of the organization. A single interface per server that results in a combined control and data plane which is needed for a fully functional OpenStack cloud. Many organizations choose to organize their cloud this way especially when port density is at a premium or the environment is simply used for testing. In production clouds separate control and data interfaces are recommended.
3.6 Types of networks in Neutron
There are two categories in Neutron to describe networks that provide connectivity to instances: • Provider networks
• Tenant networks
• Provider networks are networks which are created by the OpenStack administrator that map directly to a physical network in the data center. Useful network types in this category include flat (untagged) and VLAN (802.1q tagged). Other network types like local and GRE are configurable options but they are rarely implemented as provider networks. • Tenant networks are networks created by users to offers connectivity between instances within a tenant network. Tenant networks are fully isolated from each other including other networks within the same tenant.
• A local network is one that is remote from other networks and nodes. Instances linked to a local network may communicate with other instances in the same network on the same compute node but they are unable to communicate with instances in the same network that reside on another host. Because of this limitation local networks are recommended for testing purposes only.
• In a flat network no VLAN tagging or other network separation takes place. In some configurations instances can
reside in the same network as host machines.
• VLAN networks are networks that utilize 802.1q tagging to separate network traffic. Instances in same VLAN are deliberated part of the same network and also are in the same layer 2 broadcast domain. Inter VLAN routing or routing between VLANs is only possible done by the use of a router.
Figure- 2.4A fully meshed GRE or VXLAN overlay network is built between all hosts.
III.SIMULATION ANALYSIS
1. Overview of Openstack Network Storage.
Figure- 3.1 Overview of Openstack Storage
2. Creating instances (Virtual Machines) on openstack.
Figure- 3.3 This shows overview of instance (VM)
3. Creating multiple networks on different layers.
Figure- 3.5 This shows the presence of different network layers present in the network
Figure- 3.7 Subnet and port details of network layer 2
4..Creating and assigning routers to different network on different network layers.
Figure- 3.9 Creating Routers and assigning them to networks
IV.RESULTS
In this topology, the external network which is used to move traffic in and out of the Setup. One interface of this ext network is connected to the virtual router and the other interface of the same virtual router is connected to the tenant network(the one which we create in our environment which is network1, network2, network3 ).
The functional of virtual router is to route the traffic from multiple tenant network to the external network (Internet).
Figure- 4.19When three VMs are used to route the traffic from the networks and helps in spreading data from one network to another network on different layers of network.
I have gone to the console of this VM and ping the google.com host address which is located somewhere outside in the internet.
By using tcpdump, I came to know that the packet is transferred and showing the echo request and echo reply and length of the packet.
V. CONCLUSION AND FUTURE WORK
By using ability to create multiple tenant networks which are ultimately the networks created using VXLAN, GRE in the virtual environment, we are able to show how these multiple isolated networks can be created and used on a single physical media. Due to this network namespace isolation benefit, you can give same IP address to multiple machines hosted on multiple tenants and they all will run efficiently without getting IP conflication.
This conclusion is the base for network multitenancy in the cloud environment which is being rapidly adopted by many cloud vendors like Openstack, Vmware, etc
REFERENCES
1. H. Wang, Q. Li, G. D’Agostino, S. Havlin, H. E. Stanley, and P. Van Mieghem, “Effect of the interconnected network structure on the epidemic threshold,” Phys. Rev. E, vol. 88, no. 2, Aug. 2013.
2. K.-M. Lee, J. Y. Kim, S. Lee, and K.-I. Goh, “Multiplex networks,” in Networks of Networks: The Last Frontier of Complexity, series
Understanding Complex Systems, G. D’Agostino and A. Scala, Eds. New York, NY, USA: Springer, 2014, pp. 53–72.
3. J. Gao, D. Li, and S. Havlin, “From a single network to a network of networks,” Nat. Sci. Rev., vol. 1, no. 3, pp. 346–356, Jul. 2014.
4. P. Pattison and S. Wasserman, “Logit models and logistic regressions for social networks: II. Multivariate relations,” British J. Math. Statist. Psychol., vol. 52 (Pt 2), pp. 169–193, Nov. 1999.
5. D. Easley and J. Kleinberg, Networks, Crowds, and Markets: Reasoning About a Highly Connected World. Cambridge, U.K.: Cambridge Univ.
Press, 2010.
6. R. Ba~nos, J. Borge-Holthoefer, and Y. Moreno, “The role of hidden influentials in the diffusion of online information cascades,” EPJ Data
Sci., vol. 2, no. 6, pp. 1–16, 2013.
7. N. T. J. Bailley, The Mathematical Theory of Infectious Diseases and its Applications. London: Charles Griffin & Company Ltd, 5a Crendon
Street, High Wycombe, Bucks HP13 6LE, 1975.
8. H. W. Hethcote, “The mathematics of infectious diseases,” SIAM Rev., vol. 42, no. 4, pp. 599–653, 2000.
9. B. Min and K. Goh. (2013, Jul.). Layer-crossing overhead and information spreading in multiplex social networks, arXiv preprint 1307.2967
11. M. Granovetter. (1978, May). Threshold models of collective behavior. Am. J. Soc. [Online]. 83(6), pp. 1420–1443. Available: http://www.jstor.org/stable/2778111
12. D. J. Watts, “A simple model of global cascades on random networks,” Proc. Nat. Acad. Sci., vol. 99, no. 9, pp. 5766–5771, Apr. 2002.
13. M. De Domenico, A. Sol_e-Ribalta, E. Cozzo, M. Kivel€a, Y. Moreno, M. A. Porter, S. G_omez, and A. Arenas, “Mathematical formulation of
multilayer networks,” Phys. Rev. X, vol. 3, no. 4, p. 041022, Dec. 2013.
14. Y. Sun, C. Liu, C.-X. Zhang, and Z.-K. Zhang, “Epidemic spreading on weighted complex networks,” Phys. Lett. A, vol. 378, no. 78, pp. 635–
640, 2014.
15. S. G_omez, A. D_ıaz-Guilera, J. G_omez-Garde~nes, C. J. P_erez-Vicente, Y. Moreno, and A. Arenas, “Diffusion dynamics on multiplex
networks,” Phys. Rev. Lett., vol. 110, no. 2, p. 028701, Jan. 2013.
16. P. Domingos and M. Richardson, “Mining the network value of customers,” in Proc. ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining,
2001, pp. 57–66.
17. J. Leskovec, A. Krause, C. Guestrin, C. Faloutsos, J. VanBriesen, and N. Glance, “Cost-effective outbreak detection in networks,” in Proc. 13th
ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining, 2007, pp. 420–429.
18. M. Eslami, H. R. Rabiee, and M. Salehi, “DNE: A method for extracting cascaded diffusion networks from social networks,” in Proc. IEEE
Conf. Soc. Comput., 2011, pp. 41–48.
19. R. Banos, J. Borge-Holthoefer, and Y. Moreno, “The role of hidden influentials in the diffusion of online information cascades,” EPJ Data Sci.,
vol. 2, no. 6, pp. 1–16, 2013.
20. Z. Li and J. Yichuan, “Cross-layers cascade in multiplex networks,” in Proc. Int. Conf. Auton. Agents Multi-Agent Syst., 2014, pp. 269–276.
21. M. Magnani and L. Rossi, “The ML-model for multi-layer social networks,” in Proc. Int. Conf. Adv. Social Netw. Anal. Mining, 2011, pp. 5–
12.
