The Future of Broadband Wireless (and the role of “awareness” in wireless Internet performance)
Carey Williamson iCORE Professor Department of Computer Science University of Calgary
October 16, 2002 1
Introduction
r It is an exciting time to be an Internet
researcher (or even a user!) r The last 10 years of Internet development have brought many advances:
m m m m m m
World Wide Web (WWW) Media streaming applications “Wi-Fi” wireless LANs Mobile computing E-Commerce, mobile commerce Pervasive/ubiquitous computing
2
October 16, 2002
October 16, 2002
3
The Wireless Web
r The emergence and convergence of these
technologies enable the “wireless Web”
m m m
the wireless classroom the wireless workplace the wireless home
r Holy grail: “anything, anytime, anywhere”
access to information (when we want it, of course!) r My iCORE mandate: design, build, test, and evaluate wireless Web infrastructures
October 16, 2002 4
Clarification
“Wireless Communications”
(the enabler)
“Wireless Internet”
(the value-added service)
October 16, 2002
=
5
Internet Protocol Stack
r Application: supporting network
m
applications and end-user services
FTP, SMTP, HTTP, DNS, NTP
Application
r Transport: end to end data transfer m TCP, UDP r Network: routing of datagrams
Transport
Network
from source to destination
m
IPv4, IPv6, BGP, RIP, routing protocols
Data Link
Physical
r Data Link: hop by hop frames,
m
channel access, flow/error control
PPP, Ethernet, IEEE 802.11b
001101011...
r Physical: raw transmission of bits
October 16, 2002 6
Pieces of the Puzzle
r Portable computing devices: no problem
r
r r r
(cell phones, PDAs, notebooks, laptops…) Wireless access: not much of a problem (BlueTooth, IEEE 802.11, 802.11b, “WiFi”, 802.11a, Pringles…) Security: still an issue, but being addressed Services: the next big growth area??? Performance transparency: providing an end-user experience that is hopefully no worse than that in traditional wired Internet desktop environments (my focus)
7
October 16, 2002
Research Theme
r Existing layered Internet protocol stack
does not lend itself well to providing optimal performance for diversity of service demands and environments r Who should bend: users or protocols? r Explore the role of “awareness” in Internet protocol performance r Identify tradeoffs, evaluate performance
October 16, 2002
8
Talk Overview
r Introduction
r Background m Emerging Wireless Trends and Technologies m The Future of Broadband Wireless
r The Role of “Awareness”
m m m
TCP 101 Motivating Examples Our Work on CATNIP
r Concluding Remarks
October 16, 2002 9
Brief History: Cellular/Wireless
r First Generation (1G): analog
(cellular voice, AMPS, RTMS, TACS, 1980’s)
r Second Generation (2G): digital
(IS-64, GSM, ISM-95, 8-32 kbps, 1990’s) 2.5G You are here (always on, UMTS, 334 kbps-2 Mbps, 2000’s)
r Third Generation (3G): broadband multimedia
October 16, 2002
10
Some Interesting Reading
r Brave New Unwired World (BNUW),
by Alex Lightman and William Rojas
r In a nutshell, the authors argue that:
m
m
2.5G is dead
3G is a waste of time (and money)
m 4G
is EVERYTHING!!!
11
October 16, 2002
Another Lightman Opinion
r “the success of a technology in the
marketplace is inversely proportional to the amount of hype associated with that technology prior to its release”
Examples:
ISDN BlueTooth 3G
Examples:
Internet, Web, napster, WiFi
October 16, 2002
12
What is 4G then?
r Culmination of wireless Internet revolution r Convergence of key emerging technologies:
Image Generation
802.11b
Storage technology Semiconductors
GPS Wearable Computers New Interfaces WIDs
IP-based Networks Satellite Backhaul NWs Molecular Engineering Wireless Services
October 16, 2002
Microprocessors Antenna Arrays RF elements
NanoTech Quantum
13
Some Challenges/Opportunities
r Ultra low-power processors:
m
pg 108: “could change the entire industry…”
r Services: m pg 76: “extension of the Internet to mobile devices…whole new range of Internet services…personalized, location-sensitive content…previously impossible or impractical”
r Awareness: m pg 221: “Location/context-aware applications… can determine and react to current physical computing context of mobile users… altering information presented to users accordingly”
October 16, 2002 14
The Future?
r Service-centric economy
r Significant shifting of economic power
r The “winner” is likely to be either Japan
(iMODE, DoCoMo) or China (Internet growth, wireless growth) r Reasons:
m
m
cooperation, encouragement, support from government on a national scale strategic alliances within and across industries
October 16, 2002
15
Talk Overview
r Introduction
r Background m Emerging Wireless Trends and Technologies m The Future of Broadband Wireless
r The Role of “Awareness”
m m m
TCP 101 Motivating Examples Our Work on CATNIP
r Concluding Remarks
October 16, 2002 16
My iCORE Research Team
r Martin Arlitt: Web performance,
r r r r
r
workload characterization Transport Qian Wu: TCP, ns-2 simulation Guangwei Bai: network traffic Network measurement and modeling Tianbo Kuang: wireless measurements, Data Link video compression, streaming media Nayden Markatchev: technical support Physical Grad Students: Mingwei Gong, Yujian Li, Kehinde Oladosu, Fang Xiao, Andreas Hirt, Abhinav Gupta, Gwen Houtzager
October 16, 2002
Application
17
Internet Protocol Stack
r Application: supporting network
m
applications and end-user services
FTP, SMTP, HTTP, DNS, NTP
Application
r Transport: end to end data transfer m TCP, UDP r Network: routing of datagrams
Transport
Network
from source to destination
m
IPv4, IPv6, BGP, RIP, routing protocols
Data Link
Physical
r Data Link: hop by hop frames,
m
channel access, flow/error control
PPP, Ethernet, IEEE 802.11b
001101011...
r Physical: raw transmission of bits
October 16, 2002 18
Viewpoint
r “Layered design is good;
layered implementation is bad” -Anon. r Good:
m m
unifying framework for describing protocols modularity, black-boxes, “plug and play” functionality, well-defined interfaces (good SE)
r Bad: m increases overhead (interface boundaries) m compromises performance (ignorance)
October 16, 2002
19
Research Theme
r Existing layered Internet protocol stack
does not lend itself well to providing optimal performance for diversity of service demands and environments r Who should bend: users or protocols? r Explore the role of “awareness” in Internet protocol performance r Identify tradeoffs, evaluate performance
October 16, 2002
20
Tutorial: TCP 101
r The Transmission Control Protocol (TCP) is r r r r
the protocol that sends your data reliably Used for email, Web, ftp, telnet, … Makes sure that data is received correctly: right data, right order, exactly once Detects and recovers from any problems that occur at the IP network layer Mechanisms for reliable data transfer: sequence numbers, acknowledgements, timers, retransmissions, flow control...
21
October 16, 2002
TCP 101 (Cont’d)
r TCP is a connection-oriented protocol
SYN SYN/ACK GET URL ACK
YOUR DATA HERE
FIN ACK
October 16, 2002
FIN/ACK
22
TCP 101 (Cont’d)
r TCP slow-start and congestion avoidance
ACK
October 16, 2002
23
TCP 101 (Cont’d)
r TCP slow-start and congestion avoidance
ACK
October 16, 2002
24
TCP 101 (Cont’d)
r TCP slow-start and congestion avoidance
ACK
October 16, 2002
25
TCP 101 (Cont’d)
r This (exponential growth) “slow start”
process continues until either of the following happens:
m
m
packet loss: after a brief recovery phase, you enter a (linear growth) “congestion avoidance” phase based on slow-start threshold found all done: terminate connection and go home
October 16, 2002
26
Simple Observation
r Consider a big file transfer download:
m
brief startup period to estimate network bandwidth; most time spent sending data at the “right rate”; small added penalty for lost packet(s)
r Consider a typical Web document transfer: m median size about 6 KB, mean about 10 KB m most time is spent in startup period; as soon as you find out the network capacity, you’re done! m if you lose a packet or two, it hurts a lot!!!
October 16, 2002 27
The Problem (Restated)
r TCP doesn’t realize this dichotomy between
optimizing throughput (the classic file transfer model) versus optimizing transfer time (the Web document download model)
r Wouldn’t it be nice if it did?
(i.e., how much data it was sending, and over what type of network)
r Some research starting to explore this...
October 16, 2002 28
Motivating Example #1
r Wireless TCP Performance Problems
Low capacity, high error rate
Wireless Access
Wired Internet
High capacity, low error rate
29
October 16, 2002
Motivating Example #1
r Solution: “wireless-aware TCP” (I-TCP,
ProxyTCP, Snoop-TCP, ...)
October 16, 2002
30
Motivating Example #2
r Multi-hop “ad hoc” networking
Janelle
October 16, 2002
Carey
31
Motivating Example #2
r Multi-hop “ad hoc” networking
Janelle
Yannis
October 16, 2002
Carey
32
Motivating Example #2
r Multi-hop “ad hoc” networking
Janelle
Yannis
October 16, 2002
Carey
33
Motivating Example #2
r Multi-hop “ad hoc” networking
Janelle
Yannis
October 16, 2002
Carey
34
Motivating Example #2
r Two interesting subproblems:
m
Dynamic ad hoc routing: node movement can disrupt the IP routing path at any time, disrupting TCP connection; yet another way to lose packets!!!; possible solution: Explicit Loss Notification (ELN)
TCP flow control: the bursty nature of TCP packet transmissions can create contention for the shared wireless channel among forwarding nodes; possible solution: rate-based flow control
35
m
October 16, 2002
Example of Our Work
r Context-Aware Transport/Network
Internet Protocol (CATNIP)
r Motivation: “Like kittens, TCP connections
are born with their eyes shut” - CLW 2002
r Research Question: How much better could
TCP perform if it knew what it was trying to accomplish (e.g., Web document transfer)?
36
October 16, 2002
Some Key Observations (I think)
r Not all packet losses are created equal
r TCP sources have relatively little control r IP routers have all the power!!!
October 16, 2002
37
Tutorial: TCP 201
r There is a beautiful way to plot and
visualize the dynamics of TCP behaviour r Called a “TCP Sequence Number Plot” r Plot packet events (data and acks) as points in 2-D space, with time on the horizontal axis, and sequence number on the vertical axis
October 16, 2002
38
SeqNum
Key: X Data Packet + Ack Packet
+ X + X + X + X + X + X + X
+ + + +
X X X X X + X X + +
Time
October 16, 2002 39
TCP 201 (Cont’d)
r What happens when a packet loss occurs? r Quiz Time... m Consider a 14-packet Web document m For simplicity, consider only a single packet loss
October 16, 2002
40
SeqNum
Key: X Data Packet + Ack Packet
+ X + X + X + X + X + X + + + +
?
X X X X X + X X + +
Time
October 16, 2002 41
SeqNum
Key: X Data Packet + Ack Packet
+ X + X + X + X + X + X + + + +
X
+
X X X X X + X X + +
Time
October 16, 2002 42
SeqNum
Key: X Data Packet + Ack Packet
X X X X X X X X + X X + + + + + +
?
X X X + + +
Time
October 16, 2002 43
SeqNum
Key: X Data Packet + Ack Packet
X X X X X X
+
X +++ + + +
X X X X X + X X + +
+ + + +
Time
October 16, 2002 44
SeqNum
Key: X Data Packet + Ack Packet
?
X + X
+
Time
October 16, 2002 45
SeqNum
Key: X Data Packet + Ack Packet X X X X X X X X X + X + ++ + X + + + + + +
Time
October 16, 2002 46
TCP 201 (Cont’d)
r Main observation:
m
“Not all packet losses are created equal”
r Losses early in the transfer have a huge
adverse impact on the transfer latency r Losses near the end of the transfer always cost at least a retransmit timeout r Losses in the middle may or may not hurt, depending on congestion window size at the time of the loss
October 16, 2002 47
The TCP Transfer “Pain Profile”
Relative Transfer Time
1
SeqNum of the Single Lost Packet
N
October 16, 2002
48
Design of CATNIP
• Can we make the TCP/IP protocols “smarter” about the specific job they are trying to do? Yes. Convey application-layer context information to the TCP and IP layers
Application
Document Size Transport Packet Priority
October 16, 2002
Network
49
Design of CATNIP (Cont’d)
• Q: What could a TCP source do differently? • A: If it knew how much data it had to send, and how far along it was already, then maybe… Rate-Based Pacing of the Last Window (RBPLW)
Early Congestion Avoidance (ECA)
Selective Packet Marking (SPM): Use the reserved high-order bit in the TCP header to convey packet priority information (high priority for the really crucial packets)
October 16, 2002
50
Design of CATNIP (Cont’d)
• Q: What could an IP router do differently? • A: If it knew which packets were the “painful” ones to lose, then the router could… CATNIP-Good: give them preferential treatment, and avoid throwing them away (if possible) when congested
CATNIP-Bad: throw them away
October 16, 2002
51
Simulation Evaluation
• Network model:
Client 1 Server 1
1.5 Mbps, 5 ms
Client 2
Server 2
RouterS
RouterC
Client 99 Client 100
Server 10
October 16, 2002
52
Simulation Evaluation (Cont’d)
• Web workload model: 100 clients, 10 different Web pages Use empirically-observed distribution to determine the size, and the number of embedded images
October 16, 2002
53
Simulation Evaluation (Cont’d)
• Factors and Levels:
Factor TCP IP Levels Reno, RBPLW, ECA, ECA+RBPLW, SPM DropTail, RED, CATNIP-Good, CATNIP-Bad, CATNIP-RED
• Performance metrics: transfer time for each Web page packet loss ratio
October 16, 2002
54
Simulation Results for DropTail Routers
Reno/ RBPLW Reno
Mean and Standard Deviation of Transfer Times
ECA
ECA/RBPLW
October 16, 2002
55
Simulation Results for CATNIP-Good Routers
Mean and Standard Deviation of Transfer Times
Reno/DropTail
SPM/Good
October 16, 2002
56
Observations
r Sources have relatively little control
r IP routers have all the power
r Adding context-awareness at the IP
routers improves both mean and standard deviation of Web page transfer times r SPM and CATNIP-Good provide most of the benefit r Advantages of CATNIP are most prominent at low levels of IP packet loss (1-5%)
October 16, 2002 57
Summary
r There seem to be performance advantages
to bending the rules regarding the Internet protocol stack layered model r The general notion of “awareness” needs to explored in a variety of contexts
m
wireless networks, ad hoc routing, TCP/IP, Web caching, mobile computing, adaptive applications, …
r Many exciting issues to explore!!
October 16, 2002 58
The Next Steps
r Putting it all together: Web + Wireless r Wireless Internet Performance Lab (UofC) r Experimental Laboratory for Internet
Systems and Applications (UofS/UofC,CFI)
r Research Collaborations: m UofC, UofS, UofA, TRLabs, CS/ECE m Nortel? HP? Cisco? Agilent? Telus Mobility?
October 16, 2002 59
The End: Question Time!
r For more information: m Email: carey@cpsc.ucalgary.ca m URL: www.cpsc.ucalgary.ca/~carey
r Many thanks to my research team and the
TeleSim Research Group at the U of C r Special thanks to iCORE, NSERC, CFI, andTelus Mobility
October 16, 2002
60