Access Network for Future Internet Deokjai Choi

Outline  Changes of Networking  Access Network Technologies -Current -Coming: Sensor Networks, WMNs, DTN -Future ?  Discussions

Changes of Networking  Environment -Trusted => Untrusted  Users -Researchers => Customers => Things  Operators -Nonprofits => Commercial  Usages -Host-oriented => Data-centric  Connectivity -E2E IP => Intermittent Connection  Application Architecture -Client-Server => P2P

New Networks and Services  Home Networks  PANs  BANs  CDN  Sensor Networks, IoT  Intelligent Things  Context Aware Services  Social Networks  Smart Phone

Access networks Q: How to connect end systems to edge router?  residential access nets  institutional access networks (school, company)  mobile access networks

What is a Access Network?  Existing World -Customer Network, Access Network, Core Network (Hierarchical structure) -Accessed by residential user, customer organization, and mobile user -Access to central server, core network for delivery packet  Current and Coming World -We do not know the structure since we are trying to design now. -Even though there have been some researches for future internet architecture in research societies, we still have not seen any concrete one yet. It would not be soon to see the one. -Even though strict hierarchical structure will be getting weak a little because of P2P and CDN etc, but the principle of hierarchical structure will remain as it is (eg. Roads)

New Project by NSF-CISE (Aug. 27, 2010 ~ ) Networking Lab, Kyung Hee University 7  Named Data Networking: Lixia Zhang(UCLA)  Technical challenges: Routing scalability, fast forwarding, trust model, network security, content protection and privacy, and communication theory  Mobility First: Dipankar Raychaudhuri (Rutgers University)  Using GDTN, tradeoffs between mobility and scalability and on opportunistic use of network resources to achieve effective communications among mobile endpoints  NEBULA: Jonathan Smith (University of Pennsylvania)  The technical challenges in creating a cloud-computing-centric architecture  eXpressive Internet Architecture: Peter Steenkiste (CMU)  Refine the interface between the network and users; analyzing the relationship between technical design decisions and public policy

Access Networks  Current Access Networks -For home: ADSL, -For Organization: T1, T3 -For mobile user: Wi-Fi, WiMAX, 3G, 4G,..  Coming Access Networks for New Services -No Change -Static or mobile Human (Home, Office) -wearable devices: through some kind of gateway -New Service style -Static or mobile Sensor Network (IoT)  New or Emerging Citizen -Intermittent connection

Sensor Networks  Sensor Networks -Will be proliferated with wide usage such as environment monitoring, surveillance monitoring, bridge and building safety monitoring etc. -Most likely, they will have a sink node. -A group of nodes is connected to the Internet through a sink node which plays a role of gateway.

Why “Real” Information is so Important? Improve Productivity Protect Health High-Confidence Transport Enhance Safety & Security Improve Food Save Resources Preventing Failures Increase Comfort Enable New Knowledge

WSN Applications  Monitoring Spaces -Env. Monitoring, Conservation biology,... -Precision agriculture, -built environment comfort & efficiency... -alarms, security, surveillance, …  Monitoring Things -automated meter reading -condition-based maintenance -disaster management -Civil infrastructure  Interactions of Space and Things -manufacturing, asset tracking, fleet & franchise -context aware computing, non-verbal communication -Assistance - home/elder care  Action and control -Optimizing processes -Automation

Canonical Sensor Network Architecture Transit Network (IP or not) Access point - Base station - Proxy Sensor Patch Patch Network Data Service Intranet/Internet (IP) Client Data Browsing and Processing Sensor Node Gateway Other information sources Sensor Node

Ways of Connecting WSN to Internet

Proxy or Gateway  Protocols for WSN are free choice.  Two types: relay or front-end  Relay -Clients may register its interests to the proxy. -Data are passed through  Front-end -The proxy proactively collects data from SN and stores them in its database, and later responds to the query.  Problems: -single point of failure -One proxy for one application; it may requires many proxy implementations.

examples of Gateway  Application gateway -Works in application layer -P2P USN Sharing (example)  DTN -Works in network layer -Bundle layer is to store and forward between networks

Overlay  TCP/IP overlay sensor network -Each sensor node implements TCP/IP. -Limited resource constraints problem. -IP/USN, 6LowPAN  Sensor Network overlay TCP/IP -Each TCP/IP node implements sensor node protocols. -How many sensor node protocols should be implemented? Could it be generalized?

Overlay - IP/USN

IP Lesson  Separate the logical communication of information from the physical links that carry the packets. -Routing -Security Token Ring Ethernet WiFi 802.3a Ethernet 10b i Ethernet 10bT 802.3y Ethernet 100bT 802.3ab Ethernet 1000bT 802.3an Ethernet 1G bT a WiFi b WiFi g WiFi n WiFi X3T9.5 FDDI Serial Modem GPRS ISDN DSL Sonet Internet Protocol (IP) Routing Transport (UDP/IP, TCP/IP) Application (Telnet, FTP, SMTP, SNMP, HTTP) Diverse Object and Data Models (HTML, XML, …) LoWPAN Internet Protocol (IP) Routing

But, …  isn’t IP too heavyweight for low-power, wireless, microcontroller based devices?  No!  6lowpan compression with high quality multihop routing -Reliability and lifetime of the best mesh -Interoperability of IP

Gateway - P2P Approach to USN Integration  Adopting P2P techniques, each USN with a gateway act as a peer  The main goal of P2P overlay is to treat the underlying heterogeneou s USNs as a single unified network, in which users can send queries without considering the details of the network  User peers communicate with gateway peers in a P2P approach [Lei Shu, SAINT 2008]

P2P USN Approach  General P2P overlay network for USN Service -If a P2P peer software is installed in sink nodes, sensor nodes, and users, all USNs can be shared by users and other USNs. -USN application service is possible without knowing its target USNs protocols.  Service Scenarios -A peer node (user) can find sensor networks which can provide sensor information it wants. -A USN can find other USN for collaboration -A USN can find a peer node (user) which needs its sensory information  Advantages -Share already deployed sensor networks and need not deploy new sensor networks for specific USN service. -Exploit various information of USNs -P2P USN becomes an infrastructure for general service providers

Sink Node Architecture DB TCP/IP ZigBee 1. Service description 2. request service 3. Sensing data 4. Clear to service Application Sink module P2P Overlay module

Sensor P2P Service for Sharing USNs  P2P USN Service Scenario -USN’s sink node or a sensor node can be a P2P node and advertize own services / information. -a P2P node can also advertize services / information it wants. -a P2P node can find a service / information it wants and ask it to peer node. -a sink node or sensor node can find a peer node (user or other USN) which wants its service / information and provide that. Sensor Sink Sensor Network Overlay Network Layer (Forwarding) KOREN (Sensor P2P Layer) Peer Node Overlay Node

P2P USN Service Scenarios  An Application server finds and gathers information.  Sensor network looks for users, if special events happen Server Sensor P2P Overlay Network Sensor Network User Internet Peer Node User Event Sensor P2P Overlay Network User Sensor Network

Unstable Connection ex: SpoVNet  Spontaneous Virtual Networks -Connecting Sensor Network Islands to the Future Internet using the SpoVNet Architecture

Motivation/Objectives  Heterogeneity of network technologies makes the controllability of complex, global communication systems difficult.  SpoVNet follows the approach of providing spontaneous communication by composing algorithms and protocols that allow self-organization in distributed systems.  Self-organizing systems are able to adapt to the given requirements and network loads flexibly, without further involvement of administrative expenditure.  The main objective of spovnets is to provide the actual arising service needs spontaneously, autonomously and adaptively

Cargo Tracking System  Today’s Cargo tracking system - Consist of GPS receiver and a mobile phone unit - Attached to the actual cargo container - allows tracking of container locations  Online monitoring tracking system - The GSM unit in current location tracking systems is not limited to the transfer of GPS coordinates, but also of other sensor information too. - To reduce costly GSM communication, Several containers can use a single GSM unit that is attached to a dedicated container. - Cost and availability of GSM communication is still problematic and only allows transmission of data at large intervals

Cargo Tracking System  However, It is not satisfying - No continuous connectivity is available, therefore disallowing online monitoring - Communication is costly, making monitoring expensive  So, they employed a new Container Monitoring Application (CMA) on top of SpoVNet that uses SNS to access sensor network islands and performed the actual communication for monitoring application.

SpoVNet Sensor Network Service and Container Monitoring Application in the SpoVNet Architecture

Future Internet Access Network Technologies: Delay Tolerant Network (for another unstable connection)

Motivation  Evolve wireless networks outside the Internet -Problems with inter-networks having operational and performance characteristics that make conventional networking approaches either unworkable or impractical. -Accommodate the mobility and limited power of future wireless devices  Examples of wireless networks outside the Internet: -Terrestrial civilian networks connecting mobile wireless devices including personal communicators, intelligent highway and remote Earth outposts. -Wireless military battlefield networks connecting troops, aircraft, satellites and sensors (on land or water) -Outer-space networks, such as the “Interplanetary communications”.

Internet Evolving Concept

Why DTNs?  Current Internet was designed for -Continuous, bidirectional end-to-end path -Short round-trips -Symmetric data rates -Low error rates  Many evolving and challenged networks do not confirm to the current Internet’s philosophy -Intermittent connectivity -Long or variable Delay -Asymmetric data rates -High error rates

DTN Concept  Build upon the extended “bundling” architecture (an end-to-end message- oriented overlay) -Proposes and alternative to the Internet TCP/IP end-to-end model. -Employs hop-by-hop storage and retransmission as a transport-layer overlay. -Provides messaging service interface (similar to electronic mail)  The wireless DTN technologies may be diverse -E.g.: RF, UWB, free-space optical, acoustic (solar or ultrasonic) technologies …

Current Internet vs. DTN Routing

Types of DTN contacts  Persistent contacts

 On-demand contacts Types of DTN contacts

 Persistent contacts  On-demand contacts  Intermittent – scheduled contacts (predicted contact) Types of DTN contacts

 Persistent contacts  On-demand contacts  Intermittent – scheduled contacts (predicted contact)  Intermittent – opportunistic contacts Types of DTN contacts

DTN Routing Approach  Probabilistic Routing -Probabilistic routing methods use nodes' past encounter records to predict their future encounter probabilities  Social-Network Based Routing -Groups frequently encountered nodes into a cluster for efficient intracommunity communication and selects nodes having frequent contacts with foreign communities for intercommunity communication.  Location-Based Routing -Location-based routing methods use previous geographical location to assist packet routing in DTNs  Inter-Landmark Routing -Selects popular places that nodes visit frequently as landmarks and divides the entire D TN area into subareas represented by landmarks

DTN Probabilistic Routing  Based on assumption that real users are not likely to move around randomly  Real users have tendency to move in a predictable fashion based on repeating behavioral patterns  Example : if a node has visited a location several times before, it is likely to visit that location again.  Example : if a pair of nodes has made contact several times, it is likely to made contact again.

DTN Probabilistic Routing  When two nodes meet, they exchange summary data which also contain the delivery predictability information  The data will be transferred to the other node if the delivery predictability is higher than current nodes  Reference Project : PROPHET (Probabilistic Routing Protocol using His tory of Encounters and Transitivity)  Reference : A. Lindgren, A. Doria, and O. Schelén, “Proba bilistic routing in intermittently connected networks,” Mobile Comput. Commun. Rev., vol. 7, no. 3, pp. 19–20, 2003.

DTN Social-Network Based Routing  Based on social networks attribute  Social networks exhibit the small world phenomenon which comes from the observation that individuals are often linked by a short chain of acquaintances  Node encounters are sufficient to build a connected relationship graph, which is a small world graph  Node encounters classified into 2 types : -Intracommunity encounters -Intercommunity encounters

DTN Social-Network Based Routing  In the example : Source S want to send message to destination D  Need to find the “bridge” which is the path connecting three clusters  In the figure, i1 have weak acquaintance ties with i2, and i3 also have weak acquaintance ties with i4  These “ties” can make a path/bridge to forward data, the connection between the clusters would not exist if there is no ties  Reference Project : SimBet Routing  Reference : E. M. Daly and M. Haahr, “Social network anal ysis for routing in disconnected delay-tolerant MANETs,” in Proc. ACM MobiHoc, 2007,pp. 32–40.

DTN Location-Based Routing  Based on notion of location distribution, which calculated using location information and frequency from node history  Upon the meeting of two nodes, our approach compares their distributions and chooses the subsequent carrier for a message bundle accordingly

DTN Location-Based Routing  Routing decision based on previous node movements with a probabilistic node meeting heuristic  The nodes’ movement patterns are reactively compared to the destination’s pattern  The probabilistic meeting score denoting of how probable it is that node and the destination node have a common movement domain  Reference Project : GeoDTN (Geographic Routing in Disruption Tolerant Networks)  Reference : J. Link, D. Schmitz, and K. Wehrle, “GeoDTN: Geographic routing in disruption tolerant networks,” in Proc. IEE E GLOBECOM, 2011, pp. 1–5.

DTN Inter-Landmark Routing  Based on combination from probabilistic routing and location-based routing  From the information of how frequent a node visit an area, landmark is selected  Each landmark, configured with a central station, will determine the route to the destination area  Each node transit on landmark will relay packet to the next landmark  This routing does not only rely on nodes that frequently visit packet's destination to forward the packet, but utilize all nodes mobility  Reference Project : DTN-FLOW  Reference : K. Chen and H. Shen, "DTN-FLOW: Inter-Land mark Data Flow for High-Throughput Routing in DTNs," IE EE/ACM TRANSACTIONS ON NETWORKING, vol. 23, no. 1, pp , 2015.

Discussions  Future Internet ? -We do not know the picture at this moment.  Access Network? -We can think still there will be need to connect small things (sensors, gadget, or mobile devices) to the NETWORKs.  Major Candidates -Sensor Networks -SpoVNET style -DTN