1

Internet Trends and the Cost of Connectivity

1st BroadSky Workshop, Lacco Ameno, Italy November 6, 2003

Will Ivancic – wivancic@grc.nasa.gov

2

Experimental Satellites (70s – early 90s) Maintain US preeminence in satellite communication Focused on Commercial Communications Satellites

Space Communications (mid 90s +) NASA Mission Focused

Earth Science Computer Information and Communications Technology

(CICT) Human Exploration and Development of Space

Aeronautics (2000 +) Capacity (Air Traffic Management) Safety (Weather and Security)

NASA Glenn’s Space Communications Program

3

Aeronautics Communications, Navigation, and Surveillance (Low

Data-Rate Requirements) Delivery via VHF and/or Satellite Command and Control requires a reliable transport protocol Current protocols

Aircraft Communications Addressing and Reporting System (ACARS)

Aeronautical Telecommunication Network (ATN) Moving Toward IPv6 (commercial standards)

Weather (Low Data-Rate Requirements) Possible use of multicasting and broadcasting protocols Low Bandwidth requirements May use non-reliable transport protocols (no feedback)

Others (Medium to High Data-Rate Requirements) Entertainment Maintenance Video Surveillance Black Box Data eBusiness (Records, manuals, maps, etc…)

4

Transmission Control Protocol (TCP) Currently the Dominant Reliable Transport Protocol in

the Internet Designed to be fair and operate over shared

infrastructure (Congestion Dominates most end-to-end links)

Slow Start (Exponential Increase) to probe for bandwidth Rate Halving when packet is lost (Multiplicative Decrease) Rate Increase by 1 packet per round trip (Additive

Increase) Parameters Affecting Throughput

Bandwidth-Delay Product Congestion Errors File Size

5

Round Trip Time (RTT) Delay US to Japan Terrestrial

20 – 100 msec GEO Satellite

550 msec (Theory) Real Systems 0.8 - 1.5 sec

802.11 Wireless Ethernet Negligible (Network Delay Dominates)

G2 / G3 Cellular ~1 Second using General Packet Radio Services

(GPRS) from T-Mobile LEO

~1 – 2.5 Seconds (Using Globalstar)

6

Performance Enhancing Proxies (PEPs)

Middleware deployed to help TCP performance over links with large bandwidth-delay products

Attempt to optimize control loops Often breaks end-to-end architecture

If so, breaks end-to-end reliability (at lease at the transport layer)

Difficulty working with Security (IPSec, Virtual Private Networks)

Internet Internet

Control Loop 1 Control Loop 3Control Loop 2

End-to-End Control Loop

PEPs

7

Reliable Transport Protocol Developments TCP Swift Start

Improves Slow Start Performance TCP Westwood

Attacks Additive Increase, Multiplicative Decrease (AIMD) problem

Cumulative Explicit Transport Error Notification Attacks Additive Increase, Multiplicative Decrease

(AIMD) problem Stream Control Transport Protocol (SCTP)

New Reliable Transport Protocol Incorporates many proposed improvements to TCP

Byte Counting Selective Acknowledgements Non-Blocking of multiple streams

8

Internet Trends Inexpensive Broadband Connectivity

Cable Modem, DSL, WiFi, G2.5/G3/G4 Always On Connectivity Peer-to-Peer networking

Symmetric Links (no longer highly asymmetric) Conversation may be initiated from outside your network!

eBusiness Web replaces paper forms (e.g. eNASA, eCoast Guard)

Network Centric Warfare Mobile Networking

Maintain connections when crossing networks IPv6 Security

Network Address Translation and Proxies can really mess things up, but are probably here to stay.

Ad Hoc Networking

9

What is Mobile-IP and Mobile Networking? Mobile IP is a routing protocol that

enables IPnodes (hosts and routers) using either IPv4 or IPv6 to seamlessly “roam" among IP subnetworks.  Supports transparency above the IP layer,

including the maintenance ofactive TCP connections and UDP port bindings.

Link Independent Supports Multi-Homing (connections to more

than one route and/or media type)

Mobile Router uses

FAPort A

NOC HA Carnival Cruise

FA

FAPortB

Ethernet Ethernet

Time

11

Military Applications

AWACS

UAV

Intelligence Control Center

Battle Group Command Center (BGCC)Battle Group Command Center (BGCC)

Home-Agentdeployed inBGCC

Foreign-Agentdeployed in UAV

Foreign-Agent deployed in Tracked Command Post Carrier.Unit deployed in vicinity of the battlefield.

Tactical data forwarded from surveillance satellites to the BGCC.

Communications link between BGCC and the Field Command Post

Mobile-Router deployed in ArmoredField Unit. Secondary communications link utilized due to lost LOS of primary.

Non-preferred-pathbecomes preferred-path

Preferred-path

Non-preferred-pathIn case of communications loss of preferred-path

Mobile-Router deployed in Armored Field Unit.

INTERNET

Z

Z

ForeignAgent

HomeAgent

HQ

Hostage House

Z

Remote Command Center

INTERNET

Z

Z

ForeignAgent

HomeAgent

HQ

Hostage House

Z

Remote Command Center

14

Securing Networks Constraints/Tools

Policy Security Policy Education Enforcement

Architecture Protocols

Must be done up front to be done well

Security• Security Bandwidth Utilization • Security Performance • Tunnels Tunnels Tunnels and more Tunnels• Performance Security

User turns OFF Security to make system usable!• Thus, we need more bandwidth to ensure

security.

PAYLOADHEADER

ORIGINAL PACKET

HEADER

VIRTUAL PRIVATE NETWORK

HEADER

ENCRYPTION AT THE NETWORK LAYER

HEADER

ENCRYPTION ON THE RF LINK

16

Conclusions Regarding Security

Security Breaks Everything At least it sometimes feels like that. “The ultimate Denial-of-Service attack.” – D.S.

Need to change policy where appropriate. Need to develop good architectures that

consider how the wireless systems and protocols operate.

If you cannot change policy or architecture, then you must change the protocol.

Possible solutions that should be investigated: Dynamic, Protocol aware firewalls and proxies.

Possibly incorporated with Authentication and Authorization.

17

Satellites and Their Competition

The Cost of Connectivity

18

RF Technologies (Mobile) Globalstar (L-Band)

Globalstar MCM-8 (Client/Server) Seatel MCM-3 (Client/Server) Qualcomm MDSS-16

Boeing Connex (Ku-Band) INMARSAT Swift 64 TrackNet™ 2.0 (Ku-Band) G2.5/G3/G4

General Packet Radio Service (GPRS) – 56 kbps 1xRTT – 110 kbps

WiFi (802.11) VHF

19

Satellites vs. The World Advantages

Broadcast / Multicast

Large Coverage Area

Physical Security Surveillance Remote Sensing Navigation

(Supplemented by Governments and Defense Agencies)

Disadvantage Cost to deploy Cost of Service Time to deploy Landing Rights

(politics) Bandwidth and

Frequency reuse Point-to-Point Costs Incremental

deployment may be difficult

High Link budgets Link Delay

20

Cost of Connectivity (Examples)

21

Deployment issues (mobile) Equipment Costs Service Cost Network Peculiarities

Network Address Translators Performance Enhancing Proxies Security Mechanisms

Packet Filtering Connection Mechanisms

Smart Card Authentication MAC and/or Static Key (manual login is unacceptable)

22

Verizon Wireless Coverage

23

T-Mobile Coverage

24

Cingular Wireless Coverage

25

GSM Coverage - TerrestrialBased on Particular Service Providers

26

Satellite Coverage

Globalstar

From SaVi

INMARSAT

27

Typical Ku-Band Coverage

28

Applications, Requirements and Costs

128 kbps550 msec RTT

11 Mbps~50 msec RTT

Refernece: Ryu MIURA and Masayuki OOD: “R&D Program on Telecom and Broadcasting System Using High Altitude Platform Stations,” Journal of the Communications Research Laboratory Vol.48 No.4 2001

Stratospheric Platforms – These Are Coming Soon –

30

High Altitude Airships (Platforms)

- Coming Soon - 500 feet long, 160

feet in diameter Volume of 5.2 million

cubic feet, about 25 times larger than the blimps seen at athletic events.

21.33 km (70,000 feet) elevation

Payload 1814 kilograms (4000 pounds)

10 kWatts power

www.lockheedmartin.com/akron/protech/aeroweb/aerostat/haa.htm

The Missile Defense Agency today (Sept 29.2003) awarded Lockheed Martin (NYSE: LMT - News) a $40 million design and risk reduction contract as the next phase of the advanced technology concept demonstration to deliver a high altitude airship (HAA) prototype in 2006

www.elec.york.ac.uk/comms/presentations/HAPsmainpres2000/HAPSmainpres.pdf

32

Comments Relative to Mobile Networking Fixed Flat-Rate pricing or die

Price per bit or connect time Not manageable Impossible to budget

Voice, Video and Data are all just bits Cost of satellite equipment and services

justifies: Development of new technologies (e.g. Ad

Hoc Networks, High Altitude Airships and Stratospheric Platforms)

Deployment of new infrastructure

33

Papers and Presentations

http://roland.grc.nasa.gov/~ivancic/papers_presentations/papers.html

orhttp://roland.grc.nasa.gov/~ivancic/

and pick “Papers and Presentations”