Download presentation
Presentation is loading. Please wait.
1
PURSUIT 2011Computer LabPURSUIT 2011Computer Lab Exploring a Framework for Developing the Future Internet: Lessons from the Y-Comm Architecture Glenford Mapp Principal Lecturer, Middlesex University
2
PURSUIT 2011Computer Lab Outline of the Talk Motivation for the work Our approach Introduction to Y-Comm Peripheral Framework Areas of Work Conclusions
3
PURSUIT 2011Computer Lab Motivation (looking backward) Originally we started with ubiquitous handover. –Fore-runner was work on the Cambridge Wireless network which looked how you do handover between different wireless systems James Scott, Leo Patanapongpibul, Pablo Vidales –YComm ended up looking at ubiquitous handover, Quality-of-Service, Security and Application environments/Service Platforms
4
PURSUIT 2011Computer Lab Lesson -1.0 Each of these are big areas in themselves –Can study each area on its own and that is what a lot of people are doing. End up with good papers but not something that you could take and build a real system We took the opposite viewpoint. –Try to describe the big picture first Develop the mechanisms but keep always keep the big picture in view
5
PURSUIT 2011Computer Lab Lesson -0.5 Don’t be foolish –Invent things only when you need to Y-Comm is not trying to invent new technologies for the sake of it –Use standards or new technologies Be futuristic –Take a guess but realize you could be wrong –Allow evolution –Be conservative
6
PURSUIT 2011Computer Lab Futuristic Assumption 1: Network Evolution The Internet will evolve in a physical sense Core of the network –Super-fast backbone (optical switching, etc) –Fast access networks (MPLS, ATM) Peripheral Wireless Networks –Errors due to fading, etc; not just congestion –Handover Consequences Degradation of end-to-end arguments
7
PURSUIT 2011Computer LabPURSUIT 2011 Internet Evolution BACKBONE ACCESS NETWORKS WIRELESS NETWORKS
8
PURSUIT 2011Computer Lab Futuristic Assumption 2: Heterogeneous Devices Devices will have more than one wireless interface. Vertical handover – switching between different network interfaces to provide seamless connectivity Vertical handover is good but it introduces a lot of QoS issues because the different wireless networks have different qualities of service
9
PURSUIT 2011Computer Lab Vertical Handover (Sideffects) Affects your connections – Some protocols react badly with respect to handover. Affects your applications –Need to think through how Quality-of-Service affects applications Encapsulate these ideas in a Framework
10
PURSUIT 2011Computer Lab Layer 5: Slow Adaptation of TCP After LAN->GPRS Handover
11
PURSUIT 2011Computer Lab Lesson 0.0 New framework – We need to control network interfaces generically –Make mobility support explicit Vertical handover can have tsunami effects –Merge network and transport services –Make QoS support explicit Provide a way for applications to negotiate with the network
12
PURSUIT 2011Computer LabPURSUIT 2011 The Complete Y-Comm Framework HARDWARE PLATFORM (MOBILE NODE) HARDWARE PLATFORM (BASE STATION) NETWORK ABSTRACTION (MOBILE NODE) NETWORK ABSTRACTION (BASE STATION) VERTICAL HANDOVER POLICY MANAGEMENT END SYSTEM TRANSPORT QOS LAYER APPLICATION ENVIRONMENTS (RE)CONFIGURATION LAYER NETWORK MANAGEMENT CORE TRANSPORT NETWORK QOS LAYER SERVICE PLATFORM CORE NETWORK PERIPHERAL NETWORK SAS NTS NAS QBS
13
PURSUIT 2011Computer Lab Y-Comm Group Middlesex University Mathematical modelling of vertical handover (TBVH) and the Stream Bundle Layer for Downward QoS (Fatema Shaikh) Transport protocol and network architecture issues in Peripheral networks (Glenford Mapp) Security (Mahdi Aiash) Mobile Services (Fragkiskos Sardis) University of Cambridge Proactive knowledge- based policy mechanisms for handover (David Cottingham) Networking issues (Jon Crowcroft)
14
PURSUIT 2011Computer Lab Y-Comm Group University of Sao Paulo (ICMC San Carlos) Ontological services for vertical handover SoHand Middleware (Edson Moreira and Renata Vanni) Core Network Management (Mario Augusto) Federal University of San Carlos Testbed based on IEEE 802.21 Loughborough University Security Framework in Y-Comm (Raphael Phan)
15
PURSUIT 2011Computer Lab This talk Can’t explain everything about Y-Comm It’s too big Concentrate on the Peripheral Network See Y-Comm Research Webpage: http://www.mdx.ac.uk/research/areas/soft ware/ycomm_research.aspxhttp://www.mdx.ac.uk/research/areas/soft ware/ycomm_research.aspx
16
PURSUIT 2011Computer LabPURSUIT 2011 The Peripheral Framework APPLICATION ENVIRONMENTS LAYER QOS LAYER END TRANSPORT SYSTEM POLICY MANAGEMENT LAYER VERTICAL HANDOVER LAYER NETWORK ABSTRACTION LAYER HARDWARE PLATFORM LAYER
17
PURSUIT 2011Computer LabPURSUIT 2011 Layer 1: Hardware Platform Layer Hardware Platform Layer –Defines the physical requirements for a particular wired or wireless technology –Expanded physical layer Includes electromagnetic spectrum Modulation and channel reservation algorithms –Incompatibility issues Two technologies may be incompatible and cannot be used simultaneously
18
PURSUIT 2011Computer LabPURSUIT 2011 Hardware Platform Layer Represented as Vertical Components 3G WLAN 802.11 WiMax 802.16 UltraWideBand
19
PURSUIT 2011Computer Lab But all this is about to change! Need to make more efficient use of the electromagnetic spectrum Cognitive Radio –A radio that is aware of and can sense its environment, learn from its environment, and adjust its operation according to some objective function
20
PURSUIT 2011Computer Lab Cognitive Radio (CR) Technology –Software Defined Radio (SDR) Wide spectrum receiver Do the processing in real-time –Intelligent Signal Processing (ISP) Allows it to detect interference and move to another part of the spectrum –Ideal cognitive Radio – Mitola Radio > 2030 Mitola radio uses CR as the physical layer of a communications model That’s why CR is part of Y-Comm
21
PURSUIT 2011Computer LabPURSUIT 2011 Cognitive Radio WIDE SPECTRUM REECIVER SOFTWARE DEFINED RADIO SPECTRUM MANAGEMENT 3GWLANWiMax INTELLIGENT SIGNAL PROCESSING UltrawideBand
22
PURSUIT 2011Computer Lab Layer 2: Network Abstraction layer Network abstraction Layer –An abstraction that allows us to define, control and manage any wireless network on a mobile host –Key issues: data path functions; data formats (Link-layer), turning features on and off –Need to generate L2 triggers when a new network is detected or when an old network is no longer detectable Build on 802.21
23
PURSUIT 2011Computer Lab 802.21: Key Services Link Layer Triggers State Change Predictive Network Initiated Network Information Available Networks Neighbor Maps Network Services Handover Commands Client Initiated Network Initiated Vertical Handovers 802.21 uses multiple services to Optimize Vertical Handovers 802.21 MIH Function Protocol and Device Hardware Applications (VoIP/RTP) Connection Management WLAN Cellular WMAN L2 Triggers and Events Information Service Mobility Management Protocols Smart Triggers Information Service Handover Messages Handover Management Handover Policy Handover Messages IEEE 802.21 IETF 802.21 Overview
24
PURSUIT 2011Computer Lab Layer 3: Vertical Handover Layer Layers that define the mechanism for vertical handover. Support for different types –Network-based –Client-based
25
PURSUIT 2011Computer Lab Client-Based Handover More scalable for heterogeneous networks –Mobile node can monitor the status of all its network interfaces via the network abstraction layer – Can take into account other factors such as the state of TCP connections Don’t want to do a handover during the start and termination of TCP connections
26
PURSUIT 2011Computer LabPURSUIT 2011 Lesson 1.0 handover is complicated HANDOVER IMPERATIVE ALTERNATIVE REACTIVE PROACTIVE KNOWLEDGE-BASED MODEL-BASED NETPREF USERPREF CONTEXT SERVICES UNANTICIPATEDANTICIPATED
27
PURSUIT 2011Computer Lab Layer 4: Policy Management layer Decides if, when and where vertical handover should occur. Must be able to deal with all cases of handover –Reactive done –Proactive – most favourable but hard
28
PURSUIT 2011Computer LabPURSUIT 2011 Reactive Policy: PROTON HANDOVER EXECUTION LAYER INPUT/OUTPUT LAYER POLICY LAYER (PONDER) WLANGPRS LAN L2 Triggers HIGHER LAYERS Interface Information
29
PURSUIT 2011Computer Lab Layer 4: Proactive Policies Proactive Policy Management –The mobile node can know or estimate the network state at a given point before it arrives at that point –Proactive Policies allow us to maximize the use of available channels provided you know the amount of time a channel will be available. –That time is known as: Time before vertical handover (TBVH) Can significantly reduce packet loss during all vertical handovers
30
PURSUIT 2011Computer Lab Layer 4: Proactive policies Proactive policies can themselves be divided into 2 types Proactive knowledge-based systems –Knowledge of which local wireless networks are operating at a given location and their strengths at that point –We also need a system to maintain the integrity, accessibility and security of that data
31
PURSUIT 2011Computer Lab Proactive Policies Knowledge-based approach Gather a database of the field strengths for each network around Cambridge Need to maintain the database and also know how the results might be affected by seasonal effects
32
PURSUIT 2011Computer Lab Knowledge-Based Policy Management (Cambridge)
33
PURSUIT 2011Computer Lab Proactive Policies – Modelling Approach (Middlesex) Using a simple mathematical model Define a radius at which handover should occur Find out how much time I have before I hit that circle, given my velocity and direction Calculate TBVH Used simulation (OPNET) Can be used in the real world as well as in simulation
34
PURSUIT 2011Computer LabPURSUIT 2011 The Model-Based Handover
35
PURSUIT 2011Computer Lab Predictive Mathematical Model for TBVH (Simple Case) Introduction of additional functionality to Base Station at network boundary (BBS). Distance between MS and BBS derived from location co-ordinates or Estimated TBVH Movement of MS under BBS coverage (upward vertical handoff)
36
PURSUIT 2011Computer Lab Simulation and Results TBVH simulation in OPNET Modeler:
37
PURSUIT 2011Computer Lab A B C S T NET A NET B NET C Combining Transport and Communications to determine the optimum handover
38
PURSUIT 2011Computer Lab A H1 C1 C2 B Y1 Z1 E1 Y2 Z2 C Y3 Z3 H3 E2 E3 S T H2 Analysis shows that it is possible to calculate these key points with some degree of accuracy
39
PURSUIT 2011Computer Lab This is getting interesting Because we can not only work out TBVH You can also calculate the amount of time a mobile node will be in a given network Try to optimize handovers when networks overlap –Depends on the velocity (affects the exit radius) and adaptation time
40
PURSUIT 2011Computer LabPURSUIT 2011 Vertical Handover POLICY MANAGEMENT LAYER DECISION HANDOVER (BASE-STATION, 3G, QOS, TBVH) GPS Location, Speed, direction Connections (QoS) VERTICAL HANDOVER LAYER ACQUIRE RESOURCES ( 3GCHAN, BASE-STATION, QOS) NETWORK MANAGEMENT LAYER Send to Mobile TOPOLOGY, RESOURCES, QoS TBVH New QoS New IP RECONFIGURABLE LAYER ACQUIRE CHANNEL (3G, BASE-STATION, QOS) NETWORK ABSTRACTION LAYER BASE-STATION CHANNEL ACQUIRED DO IT NETWORK ABSTRACTION LAYER DATA CHANNNEL = 3G 3G=ACTIVE WLAN=PASSIVE WiMAX= PASSIVE 3GWLANWiMax 3GWLANWiMax Done DO IT L2 events Media Info CORE NETWORK Polling
41
PURSUIT 2011Computer Lab Lessons 2.0: We need to move to Location- based Information infrastructure Leads to better handover –Leads to a much better use of network infrastructure –It needs to tell the mobile node about individual networks Power of the transmitter, where the access points are located. – It also needs to know about the relationships between the individual networks.
42
PURSUIT 2011Computer Lab Layer 5: End User Transport System Specifies how data is routed to individual hosts and transport protocols for error correction, reliability and Quality-of-Service requirements –Encompasses Layer 3 and Layer 4 in the OSI model Different approaches –Keep the same protocols as in the core network –Keep TCP/IP, but modify TCP –Don’t modify TCP but try to get it to respond more quickly to network outages –Try a completely new protocol suite
43
PURSUIT 2011Computer Lab Layer 5: The case for a new transport Infrastructure A new transport system could be more suited for wireless networking Do all machines have to have an IP address to use the Internet? No.. Look at Network Address Translation (NAT) Translation is done between a private address and port to a global address and port at the NAT server
44
PURSUIT 2011Computer Lab Layer 5: Continued A global IP address in the case of NAT is really being used as an endpoint in the core network So we can use another network scheme in the peripheral network once we can specify how we map it to TCP/IP or UDP/IP in the core network
45
PURSUIT 2011Computer Lab Y-Comm’s view of the Future PERIPHERAL WIRELESS NETWORK CORE NETWORK PERIPHERAL WIRELESS NETWORK Core Endpoints In Access Network QoS, Secure Connection
46
PURSUIT 2011Computer Lab View corresponds with other ideas Faster LAN systems –Optimize DNS, streaming video from local caches Integrate protocol with applications –Tuneable transport –Remove the user/kernel limitations Use as a local signalling protocol –Encapsulates TCP packets
47
PURSUIT 2011Computer LabPURSUIT 2011 Simple Protocol PK_TYPEFlags CHKSUM TOTAL_LENPBLOCKTBLOCK PRICB SYNC_NO WINDOW_SIZE MESS_SEQ_NO MESS_ACK_NO DEST_ID SRC_ID
48
PURSUIT 2011Computer Lab The Simple Protocol DEST_ID (16) – identifying remote end SRC_ID (16) – from source end PK_TYPE (4) - packet type PRI (2) - supports 4 priority levels CB (2) – supports ECN CHKSUM (16) – sixteen bit checksum TOTAL_LEN (16) – total packet length PBLOCK (8) – the present block TBLOCK (8) – the total number of blocks MESS_SEQ_NO (16) – last message sent MESS_ACK_NO (16) – last message received SYNC_NO (8) – the last ACK received WINDOW_SIZE (24) – the window size
49
PURSUIT 2011Computer Lab Lesson 3.0: Local support is now essential to the Future Internet Presence of heterogeneous networks Support for mobility Need for local signalling Need for local transports –Don’t need to change TCP Need for local knowledge Can’t ignore that any more
50
PURSUIT 2011Computer Lab Heterogeneous Networking and Multi-homing Because devices will have several interfaces, there will be multi-homing issues Solutions such as Mobile IP which depend on Home and Care-of-Addresses are not scalable SCTP helps but works on the transport level only Need to look at the network addressing scheme
51
PURSUIT 2011Computer Lab Does IPv6 help? Look at the IPv6 Address Format
52
PURSUIT 2011Computer Lab Advantages of IPv6 Uses a global Interface_ID so it is very easy to construct a new address using auto-configuration –So I don’t need a foreign agent –Since the Interface_ID does not change if I do a horizontal handover, it is easier for the network infrastructure to work out what is going on Does not help with vertical handover
53
PURSUIT 2011Computer Lab Pressure for Change Mobile Systems – Efficient Vertical Handover (Y-Comm) Multi-path TCP –Allowing TCP to use all the device interfaces to transfer data IETF, Jon Crowcroft, Cambridge Pressure to optimize network resources –Switches and routers have limited memory, etc. Internet of Things –Networking should be based around devices themselves not the interfaces they use
54
PURSUIT 2011Computer Lab Different Approach Split the IP address into 2 distinct parts –Node_ID : identifies the device Independent of the number of Interfaces the device has Given when the device was created; burnt into device; Recycled when the device is no longer used –Location_ID: identifies which network the device is using. A multi-homed device will have several Location_IDs
55
PURSUIT 2011Computer Lab Proposed IP Address Format LOCATION_ID NODE_ID 127 0 63
56
PURSUIT 2011Computer Lab History behind this First proposed by LIN Group in Japan (2000- 2001) Some work done at AT&T Labs on supporting this paradigm –Mapp and Fraser (2001) Led to the development of EUI-64 –Evolve the Ethernet Mac Address into a Node_ID Current work on ILNP lead by Saleem Bhatti from St Andrews
57
PURSUIT 2011Computer Lab Why is this good? Node_ID allows the networking infrastructure to know that network interfaces are co-located Optimize network resources –In the core network we could use the Node_ID as a general guide to where the device is located. Use the Location_ID to choose which particular local network to use to forward the packet to that device –Refinement of HMIPv6
58
PURSUIT 2011Computer Lab Additional Help is needed Some observations –It would be good to know that the device is stationary in a given network Cache the Location_ID directly in the routing table –Good for servers as they are stationary –Reduce the visibility of servers Need to reduce DDoS attacks –Support of using different interfaces Multicast, broadcast, local any-cast, etc.
59
PURSUIT 2011Computer Lab So we are proposing to modify to address format LOCATION_IDNODE_ID 0127 63 NODE_ID LOCATION_ID 127055 63 NET ADMIN
60
PURSUIT 2011Computer Lab NODE_ID = Global Node Identifier (EUI-64) 063 NETADMIN = INF M SF S 08 LOCATION_ID = SUBNET 0 55 GLOBAL ROUTING PREFIX 39 NEW STRUCTURES
61
PURSUIT 2011Computer Lab SF- Scope Field Introduced by the Y-Comm Group –Ring-based Security model Aimed at protecting servers by reducing access to them Servers act within a defined network scope –Only entities within that scope can access the server –If you are outside the scope, your packet is blocked by the network infrastructure
62
PURSUIT 2011Computer Lab Scope Field – 2 bits 00 - Only processes on the same machine can use this server Location_ID must be the loop-back address 01 - Only processes on the same LAN can access the server Location_ID must be the same LAN as server 10 – Only processes on the same Admin Domain can access the server Location_ID must be site-address 11 – Server is globally accessible
63
PURSUIT 2011Computer Lab Other bits in the NetAdmin Field S – indicates that the device is stationary with regard to a given network –Physically stationary – Servers –Relatively stationary as determined by a location system M – indicates that Node_ID is being used as a multicast address, this allows multicast packets to be sent on network
64
PURSUIT 2011Computer Lab Interface_IDs can make a comeback Interface_IDs banished from the New address format –Means we will have to map the Location_ID to the Interface_ID before we could send a packet on a given network. Slow performance Other observation –Bluetooth and other short range systems do not need a Location_ID (network) and Interface_ID might be better
65
PURSUIT 2011Computer Lab Re-Introduce it by indirectly INF field (4 bits) An interface can be uniquely represented within the address format by using an interface number –Controlled by the device. Map interface # to MAC address Give special meaning to certain values –0 – any-cast – packet is sent on any available network interface of the mobile device –0xF – broadcast packet is sent on all available network interfaces of the mobile device –0x1 – called the primary interface
66
PURSUIT 2011Computer Lab INF Support for pseudo/virtual interfaces –Add features such as security OpenSSL, FreeSwan, provides IPSec services Applications can query interfaces on a device –Master locator –Map interfaces #s to Quality-of-Service (QoS) parameters Bluetooth and other point-to-point system could use the INF only and not bother with the Location_ID
67
PURSUIT 2011Computer Lab Lesson 4.0: Can’t ignore multi- homing any longer Crucial for heterogeneous networking Need to concentrate on devices and not their interfaces Need to think about keeping it flexible and efficient Urgent
68
PURSUIT 2011Computer Lab Layer 6: QoS Layer QoS is the most dynamically changing component in heterogeneous networking Applications running on heterogeneous devices need support to handle this Two Concepts of QoS –Downward QoS –Upward QoS
69
PURSUIT 2011Computer Lab Layer 6: Downward QoS Mainly to support legacy applications The application specifies a minimum QoS and the QoS layer does the mapping between the QoS that the application requires and the QoS that is currently available - but is dynamically changing
70
PURSUIT 2011Computer Lab Layer 6: Upward QoS For applications that should adapt to changes in QoS, e.g. Multimedia services, etc –The QoS layer therefore signals the application using an event mechanism to indicate changes in the available QoS –Applications can specify routines that will be called when the events occur Similar to the X Window System
71
PURSUIT 2011Computer LabPURSUIT 2011 QoS Layer APPLICATION LAYER QOS LAYER END TRANSPORT LAYER POLICY MANAGEMENT LAYER DOWNWARD QOS APPLICATION LAYER QOS LAYER END TRANSPORT LAYER POLICY MANAGEMENT LAYER UPWARD QOS
72
PURSUIT 2011Computer Lab Lesson 5.0 Make QoS the real interface to the network Applications should not really have to know about TCP or any other protocol Replace transport protocols on the fly Allow applications to adapt if they can
73
PURSUIT 2011Computer Lab Last area: Security Security is part of Quality-of-Service QoSS – specify security as part of QoS –Changes in security are regarded as changes in the QoS Security needs to be on different levels and involves both Core and Peripheral Frameworks Allow several layers
74
PURSUIT 2011Computer LabPURSUIT 2011 The Y-Comm Framework showing its Security Levels- New Security Level HARDWARE PLATFORM (MOBILE NODE) HARDWARE PLATFORM (BASE STATION) NETWORK ABSTRACTION (MOBILE NODE) NETWORK ABSTRACTION (BASE STATION) VERTICAL HANDOVER POLICY MANAGEMENT END SYSTEM TRANSPORT QOS LAYER APPLICATION ENVIRONMENTS (RE)CONFIGURATION LAYER NETWORK MANAGEMENT CORE TRANSPORT NETWORK QOS LAYER SERVICE PLATFORM CORE NETWORK PERIPHERAL NETWORK SAS NTS NAS QBS
75
PURSUIT 2011Computer Lab Security in Y-Comm Network Architecture Security (NAS) Security dealing with the deployment and management of different wireless technology Managed from the Policy and Management layers Network Transport Security (NTS) –Security dealing with end-to-end transport through Y-Comm Done at Layer 5, NAT, IPSec, etc QoS Based Security (QBS) –Look at QoS Issues Looks at SLAs, prevents overloading Looks at Denial of Service Attacks Service and Application Security (SAS) –Security dealing with running applications and deploying services AAAC, ACLs, User-based security
76
PURSUIT 2011Computer Lab Lesson 6: Security needs to consider the effect of an open Architecture Security needs to be part of the communications system Also needs to be part of QoS Security is about protecting entities and not just data –Concept of Security models Connection Ring-based Vertical Handover
77
PURSUIT 2011Computer Lab Lesson 7: New business models Y-Comm is doomed to failure No academic funding –Not about TCP/IP –Not interested in tinkering around the edges No industry funding –Y-Comm will destroy the revenue streams of current mobile providers –They will not put money in unless we show them how they are going to make money
78
PURSUIT 2011Computer Lab Any Questions?
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.