Status of Next Generation Cellular and Wireless Local Area Networks and Current

Research Activities

Mohsen GuizaniComputer Science DepartmentWestern Michigan University

mguizani@cs.wmich.edu

Western Michigan University

• WMU is located in Kalamazoo, Michigan• WMU is one of 15 Michigan state schools• WMU has more than 28,000 students• The Computer Science is home to about

400 students• CS has 18 faculty members, 5 full

professors, 7 associate professors, and 6 assistant professors.

Outline Introduction Cellular Coverage in the United States Current Problems in the Telecommunications

Industry Review of Cellular Technologies Wi-Fi: Competing or Complementary

Technology? The Future Current Research Activities Conclusions

Current Research Activities Research Goal 1x EV-DV Architecture Resource Allocations Techniques Cross Layer Design Overview Intelligent Network QoS Protocols Intelligent Network QoS Validation Protocol Wireless QoS Based Routing Protocol

Introduction A combination of factors has led to the current

wireless situation in the US, which is rather poor in many respects Rapid technological change Rapid change in way people use technology Poor business and investment decisions Unrealistic expectations for new technologies Competition on features and packages rather

than underlying infrastructure More thinking and intelligent decision making in

future should enable vastly improved wireless service

Cellular Coverage in the US: Reason for Poor Coverage

Coverage is similar (often poor) because all providers use the same antenna towers

Much of the engineering behind tower placement is done in the old days of 3 watt cell phones at 800 MHz in cars with external antenna; in this day and age, the is much lower-powered units inside buildings or cars with no external antennas

NIMBY (“not in my backyard”) syndrome: Wealthy neighborhoods refuse to allow unsightly antenna towers

Cellular Coverage in the US (Continued) Call by one of the authors from Baltimore, MD to

Washington DC Dulles International Airport interrupted seven times due to coverage gaps—partly ascribed to the fact that there are five major cellular providers each of which has to build an entire network

The Yankee Group estimates that it would take $50B to $100B to bring cellular system up to snuff Carriers do not have that kind of money Would not solve political problems

Convenience trumps service quality Relatively few people have abandoned landline

phones

Cellular Coverage in the US (Continued)

0

5

10

15

20

25

30

35

40

45

50

MCI Pre-Bankruptcy

$41.0

MCI Post-Bankruptcy

$5.5 *

Verizon

$45.4

Sprint

$19.2

SBC

$17.9

Qwest

$17.5BellSouth

$15.0

AT&T

$14.4

0

5

10

15

20

25

30

35

40

45

50

MCI Pre-Bankruptcy

$41.0

MCI Post-Bankruptcy

$5.5 *

Verizon

$45.4

Sprint

$19.2

SBC

$17.9

Qwest

$17.5BellSouth

$15.0

AT&T

$14.4

Source: Wall Street Journal,

4/15/04

Cost Constraints Minimal revenue per minute of air time

Brutal competition Availability of free airtime and long distance

packages No “killer app” has ever materialized

Not cameras and ability to send photos People want dependable voice communications

Cellular phones unsuitable as wireless modems Promoters did not consider human factors E-mail already well-served by dedicated devices

such as the popular Blackberry by RIM

Cost Constraints (cont.) Access to the Internet is done while at

rest Coverage problems would interrupt most

operations if in motion Cannot really do anything while driving or

walking Screen is too small

Competing technologies such as Wi-Fi are much better

Problems

Problem Reasons

Telecommunications firms are deeply in debt (or out of business) because of

two miscalculations stemming from over-estimating revenue

potential of fiber optics

Poor understanding of human factors—how quickly people will change their way of doing things

The difficulty of solving the “last mile” problem—aging copper plant; coaxial system designed for one-way transmission

Technology provided capacity far in excess of

users to absorb

$100M will buy a fiber from US to UK With WDM technology this could be >2

Tbits per second—few users

No “killer app” has materialized

The only one is sight is HDTV (video on demand)—may well happen, as HDTV is slowly being adopted; however widespread demand is still 3 to 5 years in the future

Will not be from low-bandwidth devices such as videophones

Problems (Continued)Problem Reasons

Telecommunications companies rested on their

laurels

Did not upgrade “last mile” Rollout of DSL was initially slow—many

areas could not be served; has limits in any case

Pressure built as CATV companies began to offer broadband

Only recently have telecommunications companies begun to move out on POE

Infatuation with technology and disregard of human factors has led

to other telecommunications

company fiascos, such as Iridium and Global Star

Proponents underestimated speed at which conventional cellular systems would be deployed

Never showed that a large base of users would pay $5 per minute—not enough Antarctic explorers, oil drilling rigs, Sahara desert trekkers, Mt. Everest climbers; village chief in third world country could not afford it

Problems (continued)Problem Reasons

Other: Narrowband Integrated Services Digital

Network (ISDN) (2B+D)

Developed in the late 1970s Supposed to be vehicle to convert telephony

worldwide in the 1980s Priced very high Few could see any real benefits—calls

connected faster, but audio quality about the same

128 Kbits per second was enormous by 300 bd standards in the 1980s—few places to dial-up then

By the time “killer app” arrived, the Internet, ISDN was wrong paradigm

Connection-oriented service (pay by minute) for connections protocol

56K modems were extremely cost-effective and not that much slower

Problems (continued)Problem Reasons

Other: DSL

technologies offer much higher

bandwidth than ISDN at lower cost for typical usage

Rollout delayed by ISDN push Now a successful service

Other: Telecommunication

s companies overbid for 3G

wireless spectrum to the tune of

billions of dollars

Supposed to provide great advantages, unfortunately based on several assumptions

Technology was proven Users had need for features it offered, such as video

1G Cellular Technology

Advanced Mobile Phone Service (AMPS) Analog Widest coverage, much wider than digital systems Phased out by 2008

Of concern to users of OnStar, which employs it—digital systems’ coverage poor by comparison

Being phased out because newer systems can support more customers per unit of bandwidth—bandwidth is most precious resource

2G: GSM, CDMA, IS-95-a, iDEN

Global System Mobile (GSM) Initially Group Speciale Mobile; renamed Global System

Mobile to give it an international flavor Combined TDMA/FDMA system Offered by AT&T, T-Mobile, and Cingular in the US Advantage: With unlocked tri-band phone, users can

have cellular service worldwide Problem: Outside the US, reciprocal agreements with US

providers expensive—$~4 per minute for airtime Better solutions: Get subsidy unlock code for phone used

in US—buy SIM card when abroad from kiosk; buy cheap tri-band phone in the US, then buy SIM card when abroad

2G (Continued)

Coded Division Multiple Access (CDMA) Offered by Sprint and Verizon in the US Verizon recently launched high-speed

data service based on Phase 1 Evolution Data Only (1xEV-DO) in Washington, DC and San Diego, CA

Can handle the largest number of users per unit BW; most economically attractive

2G (continued)

Time Division Multiple Access (TDMA) Use declining Offered by AT&T and Cingular in the US

Integrated Digital Enhanced Network (iDEN) Developed by Motorola Based on TDMA Offered by Nextel in the US Likely to be phased out in favor of CDMA-2000

2.5G

So-called 2.5 or 3rd generation wireless technologies unlikely to be profitable, especially given prices paid for spectrum Main thrust is higher speed data Cannot compete with Wi-Fi Nextel is planning to bypass altogether What is needed is data rate of >2 Mbps

Beyond Various generations of cellular telephony more important

to providers than users Maximize revenue per unit bw Users care more about features, cost, dependability Many features being pushed are of dubious value

Multimedia Messaging Service (MMS) Short Messaging Service (SMS) Walkie-talkie feature

Reduces connect time to ~2 second versus 15 second dial time

Games Downloadable ring tones Replaceable covers

Wi-Fi: Is It Really a Good Idea?

IEEE 802.11b Caught on very fast; manufacturers

incorporate Wi-Fi chips in laptops; hopes are that this will be the new “killer app”

Wireless LAN equipment sales have been growing—Gartner Group says 2002 spending on all vendors is ~$2.3B; end-user spending increasing by about 50 percent for the last two years

Wi-Fi: Security Issues Algorithm is used, Wired Equivalent

Privacy (WEP) discredited Encryption key length too short Initialization vector implementation flawed Scheme can be cracked quickly Successor, WPA, is patch—not a fix

Vulnerable to broadband jamming, unless it uses frequency hopping as does Bluetooth

Wi-Fi: Security Issues(Continued)

Users do not seem to care 70 percent of installations have not even

implemented what little security measures there are

Incompatibilities among vendor equipment mean that Wi-Fi hot spots must implement lowest common denominator, i.e., no security

Wi-Fi user sitting next to “me” at Starbuck’s can intercept all transmissions to/from my computer Doctor files in “my” computer Impersonate “me” after “I” have logged off

Wi-Fi: Security Issues(Concluded)

Lack of scalability PKI has not provided desired solution Efficiently and rapidly propagating

information about revoked encryption keys through large networks

Problem of where to store private or secret key safely in a manner that hacking cannot compromise

Smart cards may be the only viable solution, but most laptops have no smart card reader Could be added through USB port

Wi-Fi: Business Model No clear business model Nobody making money off of Wi-Fi

Not a cost center, but a gimmick to attract customers

Issue of illegal use of Wi-Fi connectivity—who is liable? Maryland homeowner recently held liable when

someone used his hot spot for an illegal act Airports and other such places look to Wi-Fi to recoup

money no longer received from pay phones Travelers unlikely to agree to open yet another

account unless all places they frequent use same account

Wi-Fi: Setup Difficulties and Network Incompatibilities Complex Windows’ network setup menus and options to

set the SSID for each hotspot provider’s Access Point Most non-technical laptop users are disinclined to do so

Technical help from kid behind counter at Starbuck’s, etc., is a losing proposition

Proliferation of different Wi-Fi hotspot providers means that users must open a separate account for each

T-Mobile account at 2,100 Starbuck’s or Kinko’s Cometa account at MacDonald’s FatPort account in Canada Surf & Sip account at Foley’s Irish pubs Toshiba account at Arizona’s Circle K stores Waypoint account at a few select hotels

Wi-Fi: User Fees and Speed Problems User fees

Disinclination of users to pay more access fees Many feel they are already paying their

Internet dues through home subscriptions Lots of free Wi-Fi access points

From businesses that want to attract customers for their main product

Speed problems Chips implementing 802.11b with WEP force

all users to speed of slowest user at the hotspot

Wi-Fi: Incompatibilities and Spectrum Shortage Incompatibilities between WEP and WPA

Problem has not received much press because commercial hotspots have not enabled either—due to vendor incompatibilities

Spectrum shortage 802.11a has more spectrum allocated to it

(which allows it to accommodate more concurrent users)—however has not yet caught on

Dual 802.11b/a access points and especially client user’s PCMCIA cards are very expensive; suffers from the same security vulnerabilities

Wi-Fi: Standards and Scalability Standards

802.11i, 802.11x, and 802.11e “standards” waiting in wings in various levels of agreement as to their final specs

Problem is that millions of deployed laptops and hotspots may make upgrade to better standards impossible

Scalability Inherently not scalable Operates in crowded unlicensed band with baby

monitors, cordless phones, Bluetooth devices, microwave ovens

Limited number of channels—3 versus 8 for 802.11a

Wi-Fi: Summary Wi-Fi has not really taken anything away

from cellular Cell phones are not as practical as wireless

modems at 3 Kbps to 8 Kbps Some CDMA systems (Sprint) encouraged

use of cell phone itself for e-mail and messaging; however not practical due to the small size of the keyboard

Human factors: do people really want to make coffee shops another extension of their office?

The Future Despite problems, wireless is here to stay

Convenience dictates that it users will demand it Problems of “last mile” access Need to set up and tear down networks quickly

Mobile ad hoc networks (MANETs) for military and for emergency responder use

Sets the stage for determining who will emerge victorious in future Go beyond solving current problems and

anticipate and solve future problems—foregoing; societal preferences, economics, scalability, and regulatory issues

The Future (Continued) Realities of wireless solutions

Must be commercialized within months Cannot hope for any regulatory protection given

in the past to telecommunications monopolies Will have to compete fiercely with other

technologies for customer dollars—and hence for survival

Three issues of importance Spectrum Technology available to address problems Socio-political issues

The Future (Continued)

Really is not a spectrum shortage Even in areas such as Washington,

DC, only about 20 percent of available cell phone spectrum used during peak hours

Real problem is more intelligent and efficient use of available spectrum

Technology Available to address Problems

Ultra-wideband Wi-Max Wireless mesh networks Smart antennae Software radios

Technology Available to address Problems (cont.)

Ultra-wideband Uses short (~1 nsec) pulses which correspond

to about 1 GHz bandwidth Such pulses with 1 W peak power and repetition

rate of 108 have average power of 100 mW spread over 1 GHz

Interference in a 1 kHz channel ~ 0.1 mW

FCC has allocated 3.1-10.6 GHz band Currently in use by satellite uplinks and downlinks

Data rates up to 500 Mbits per second can be accommodated versus 700 kbps for Bluetooth

Technology Available to address Problems (cont.)Ultra-wideband (Concluded) UBW likely to become

standard of choice for home networks

IEEE standard is 802.15.3a Uses TDMA Wireless Personal Area

Network (WPAN) 245 devices up to 90 m Data rates 11 – 55 Mbps,

declining with distance AES encryption Discussions now about

dividing

Expected shipments of UWB equipment

Technology Available to address Problems (cont.)

Wi-Max Another emerging technology

Intended for distances up to 50 km at data rates up to 70 Mbps

Intended to provide broadband service to replace “last mile” where this is not cost-effective with conventional technology

May also take up some of the functions of Wi-Fi

Technology Available to address Problems (cont.)Wireless mesh networks Low-powered systems that pass messages from node to

node on their way to their destination, not unlike what Internet nodes do with e-mail and other Transmission Control Protocol/Internet Protocol (TCP/IP) traffic

Any one node’s RF power output needs to be no more than what is required to close the link to the next nearest nodes

Redundant paths enhance the likelihood of end-to-end message integrity

Inherent is frequency reuse Similar to old Ricochet network which went bankrupt

because high costs of installation could not be recouped with small base of users

Technology Available to address Problems (cont.)Smart antennae Two stations communicating by wireless have

absolutely no excuse for using omni-directional antennas

If each end could beam all of its RF energy towards the direction of the intended receiver, the RF spectrum would experience a massive increase in availability with no new frequency allocations

Beam forming can be computer-controlled for adaptive beam forming

In case of cellular base stations can be fast enough to accommodate vehicular users

In case of Wi-Fi can extend range; SF startup, Vivato, working on 128 beam implementation

Technology Available to address Problems (cont.)

Software radios Software-configurable cell phones To handle multiple systems, also

Wi-Fi Eliminate need to buy new cell

phones every year or so

Socio-Political Issues Diverging international standards—China adopting

its own wireless LAN standard, basically Wi-Fi with improved security

Ad hoc implementations—Some locations installing their own area-wide Wi-Fi to deal with problem of multiple accounts (Cerritos, CA)

Voice over IP Currently a major trend, or at least major hyped

trend Promises many benefits But many legal and regulatory issues unresolved,

especially related to emergency response and USF

Conclusion Rate of change in telecommunications has been

unprecedented International cellular and wireless LAN industries have

had two decades of gross miscalculations Multibillion dollar bankruptcies Endless miles of unused fiber optic cables Digital cellular coverage in the US which is poor even

by third world country standards Wireless LAN standards whose lack of security has

been an embarrassment Hodge-podge of mutually incompatible cellular

standards

Conclusion (continued) Industry now has the opportunity to

plan wisely ahead Forego the short-term gimmickry of

downloadable ringing tones and designer-face-plates

Use US technological prowess in evolving technologies such as software radios, ultra-wideband, and smart antennas to forge standards that will with-stand the test of time and of consumer acceptance

Current Research Activities Research Goal 1x EV-DV Architecture Resource Allocations Techniques Cross Layer Design Overview Intelligent Network QoS Protocols Intelligent Network QoS Validation Protocol Wireless QoS Based Routing Protocol

Motivations

High bit-rate applications (www, file transfer, full motion video) impose strong requirements/needs on the system capacity

Studies confirm a productive gain of between 7-8 hours a week when business users are equipped with mobile PCs and wireless access.

All-IP applications: end to end packet-switched network

Goals To develop a new dynamic and

intelligent resource allocation technique for optimizing the average throughput of the wireless system.

Maximize the spectral efficiency and the number of users supported.

Develop QoS based protocol in the upper layer to assure the level of service required.

Block Diagram

Physical Layer

MAC Layer

MAC/Network QoS Mapping Layer

Intelligent Network QoS Validation Protocol Engine

Res. Alloc.

Network QoS

Competing technologies

CDMA Family

cdmaOne - IS-95A (2G) - IS-95B (2.5G)

CDMA 2000 1x (3G) 2000 CDMA 2000 3x MC (3G) 2001 1xED-DO (3G) 2002 1xEV-DV (3.5G) 2003

IS-95A

2G – 1995 Upto 14.4 kbps data rates Used exclusively for circuit-switched

voice Used Convolutional channel coding Used BPSK (fixed) modulation technique * BPSK: Binary Phase Shift Keying

IS-95B

2.5G – 1999 MAC layer enhanced over IS-95A Up to 115 kbps data rates (64 kbps) Up to 8 forward or reverse code channels

can be simultaneously assigned to a MSU using Walsh codes and PN sequence masks

Code channels are transmitted at full data rates during a data burst.

Used Convolutional channel coding Used BPSK modulation technique

CDMA 2000 1x 3G – 2000 Up to 307 kbps data rates (144 kbps) Q-PCH enables to monitor F-CCCH and Paging

Channel => improve battery life Radio Configurations (RC) => additional data

rates Quality and Erasure indicator bit (QIB and EIB)

on the reverse power control sub-channel. Code channels are transmitted at full data

rates during a data burst. Used Convolutional and Turbo channel coding Used QPSK modulation technique

CDMA 2000 3x MC

3G – 2001 Up to 2 Mbps data rates Using 3 standard 1.25 MHz Chs

within a 5 MHz band Used Convolutional and Turbo

channel coding Used QPSK modulation technique

1xEV-DO 3G – 2002 1st Evolution phase of CDMA2000 Up to 2.4 Mbps data rates No backward-compatibility with CDMA 2000 2 inter-operable modes: 1x and 1xEV modes Adaptive Rate Operation with respect to

channel conditions Adaptive Modulation and Coding (AMC) Macro diversity via radio selection Always-on operation of 1xEV-DO terminals in

the active state Multi-level modulation format (QPSK, 8-PSK, 16-

QAM)

1xEV-DV 3.5G 2003 Forward peak data rate: 3.072 Mbps Reverse peak data rate: 451.2 kbps 3 new Chs to the forward link for the packet data

operation (F-PDCH, F-PDCCH0, F-PDCCH1) 3 new Chs to the reverse link to support operation

of F-PDCH (R-RICH, R-CQICH, R-ACKCH) Adaptive Modulation and Coding on the forward link

in real time to adapt to the RF environment (QPSK, 8-PSK, 16-QAM)

Variable RF frame duration (1.25, 2.5 and 5 ms) Fast selection of base station to serve forward link No soft handoff on F-PDCH or F-PDCCH0 and F-

PDCCH1

Tracing the DR Evolution

14.4 115307

20002400

3072

0

500

1000

1500

2000

2500

3000

3500

Data Rates Evolution

Data Rate (kbps)

IS-95A IS-95B CDMA2000 1x CDMA2000 3x 1xEV-DO 1xEV-DV

CDMA Evolution Path

1xEV-DV Architecture

Logical and Physical Channels

Physical Layer Interface Control Information

MAC Layer User’s control Bearer Data

Other Layers No new service interfaces

Forward Packet Data Channel Traffic channel combinations

operate in both mixed voice and data services and data-only services in the forward and reverse links.

New Physical Channels

Forward Link Traffic Channel

F-PDCH Control Channel

F-PDCCH

Reverse Link Control Channel

R-ACKCH RCQICH

Adaptive Modulation and Coding

Adaptive Modulation and Coding

Base Station (Tx)Modulation and Coding Scheme

Mobile Station (Rx)Channel Quality

Reverse Link Feedback (R-CQICH)

CHANNEL

The base station assigns users the best modulation and coding rate for the instantaneous channel conditions (SINR).

Adaptive Modulation and Coding

Provides higher data rate services by varying The RF frame duration (1.25, 2.5 or 5

milliseconds) The number of bits per RF frame (between

408 and 3864 bits) The coding algorithm

QPSK (Quadrature Phase Shift Keying) 8-PSK (8-states Phase Shift Keying) 16-QAM (16-state Quadrature Amplitude

Modulation) .

F-PDCH Data Rates Data rates depending on F-PDCH packet size

and RF frame duration. The RF frame duration

“Number of Slots per Sub-packet” (1 slot = 1.25 ms)

Hybrid ARQ Automatic Repeat reQuest (ARQ)

Immigrates from MAC layer to Physical layer for improving performance

A mechanism supporting retransmission of frames received in error

Hybrid ARQ Chase combining, each retransmission

repeats the first transmission or part of it. Incremental redundancy (IR), each

retransmission provides new code bits from the mother code to build a lower rate code

AMC and hybrid ARQ On a single carrier, 1xEV-DV can

efficiently serve both data and legacy services (e.g., voice) by combining of Fast AMC and Hybrid ARQ Fast AMC is a link adaptation scheme where

the base station assigns users the best modulation and coding rate for the instantaneous channel conditions.

Hybrid ARQ improves throughput and enables fast AMC by making the initial modulation and code rate selection process tolerant to selection error.

F-CPCCH

R-PICH

F-PICH

(F-DCCH/FCH/SCH)

(R-DCCH/FCH/SCH)

F-CPCCH

F-PICH

R-PICH

R-CQICH

F-PDCCH

F-PDCH

R-ACKCH

(F-DCCH/FCH/SCH)

(R-DCCH/FCH/SCH)

BTS 1 BTS 2MOBILE

Cell Selection

The mobile station selects one base station from its active set

The selection based on the RF quality measured (SINR) by the mobile station

Flexible TDM/CDM Multiplexing 1xEV-DV was designed to support all

services Services that use large packets Services that use small packets

To reach the goal, TDM and CDM are included into the 1xEV-DV specifications

TDM/CDM multiplexing allows the selection of both the number of timeslots and the number of Walsh codes allocated to a user.

TDM/CDM

The TDM/CDM in 1xEV-DV system maximizes system throughput by providing optimal modulation and coding rate assignments to all services while maintaining frame fill efficiency.

A small packet may receive a few of the Walsh codes, and the remaining Walsh codes can be used by another user, improving overall system capacity

WasteUsed by other

traffic

TDM TDM/CDM

Required Required

Cod

e S

pace

Frame Duration

Modulation and Coding Schemes (MCS)

Rate (kbps) Slots Per Packet

Packet size (Bits)

Turbo Code Rate

Modulation Effective Code Rate

38.4 16 1,024 1/5 QPSK 1/48 76.8 8 1,024 1/5 QPSK 1/24 153.6 4 1,024 1/5 QPSK 1/12 307.2 2 1,024 1/5 QPSK 1/6 614.4 1 1,024 1/3 QPSK 1/3 307.2 4 2,048 1/3 QPSK 16/99 614.4 2 2,048 1/3 QPSK 16/49 1,228.8 1 2,048 1/3 QPSK 2/3 921.6 2 3,072 1/3 8-PSK 16/49 1,843.2 1 3,072 1/3 8-PSK 2/3 1,228.8 2 4,096 1/3 16-QAM 16/49 2,457.6 1 4,096 1/3 16-QAM 2/3

AMC Fixed Threshold Method

AMC has a set of n MCS levels

MCS set has a corresponding throughput vs. av. Channel SINR denoted by

These throughput values can be graphically represented, where the curves intersect with each other.

SINR at intersection points are threshold values, denoted by

}1,...,0),({ niTi

},,...,,{ 110 nn

},...,{ 10 nMM

AMC Fixed Threshold Method These threshold points partition the

range of SINR into n regions, denoted by

The kth MCS, namely Mk is assigned to the region

if the following condition is satisfied

1,...,0 ),[ 1 niforii

),[ 1ii

).,[,),()( 1 iijk kjTT

AMC Fixed Threshold Method

With this corresponding between the MCS’s and the average SINR, Mk is selected for the next frame if the average channel SINR in the current frame lies in the region ),[ 1ii

Channel Estimate

SINR

γi

MCSi

}

Threshold values, fixed

Disadvantages of TM Error in the estimation of average

channel SINR can cause inappropriate selection of MCS resulting in a degradation of the performance

The threshold values associated with the MCSs are not jointly optimized based on the overall stochastic behavior of the users’ SINR degrade the efficiency of the overall system resources.

Optimized Method

The threshold values associated with the MCSs are jointly optimized based on the overall stochastic behavior of the users’ SINRs The goal is:

Higher overall throughput

Channel Estimate

SINR

γi

MCSi

}

Threshold values, optimized

Percentage of users served by a MCSi

P1 + P2 + + PN = 1 The SINR is a random variable (r.v.)

achieved by an arbitrary user at a given instant

We prove ordinarily that Pi is a discrete random function that is dependent on the users’ joint SINR cumulative distribution function (CDF) and data rate granularity (N).

)},,[Pr{ 1 iii SINRSINRSINRP

Throughput Optimization Consider the event {SINR x} where x is a

real number in the interval [0,). We write the probability of this event as

The function F(x) is the CDF of the r.v. SINR. In

our case, F(0) ≡ F() = 0 and F(SINRN+1) ≡ F() = 1. Thus, we can rewrite Pi as

)Pr()Pr()Pr()Pr(

)Pr()Pr()Pr(

11

11

iiii

iiiii

SINRSINRSINRSINRSINRSINRSINRSINR

SINRSINRSINRSINRSINRSINRSINRP

.0),Pr()( xxSINRxF

In terms of discrete CDF functions, Pi is expressed as

)Pr()Pr()()( 11 iiiii SINRSINRSINRSINRSINRFSINRFP

Throughput Optimization

SINRi thresholds:

,/TNDRRii CpCi

,ii CpC

CHIP

ip NDR

N

R

WG

iii PLTCC RRR /

CHIPPLi

CiPTCiiobi NR

NDRNESINR

/

SINRi thresholds for variable bit rates: The SINRi threshold associated with a MCSi is determined by

p

iobi G

NESINR

)/(

Bit rate can be calculated

GP can be calculated

RCi is given by

Throughput calculations:

N

iiiii

PL

CTCTSP SINRSINRSINRSINRSINRFSINRFR

NR

T

ND

i

ii

111 )Pr()Pr()()(

,TSii NR

,_ iiieff P

i represents the throughput that can be transmitted by a base station

Let also eff_i represents the effective throughput that can be received by the users who can achieve an SINR in the range [SINRi, SINRi+1)

Simulation Model

Assume there are M possible users’ realizations over a certain period of time, then P is a member in an M-size set {Pj: j = 1, 2, … , M}.

We consider 19 3-sector cells located on a hexagonal grid and used the SINR calculations as shown

1 2

34

5

6 7

8

9

101112

13

14

15

16 17 18

19

Effective throughput

Effective aggregate throughput for 100 different realizations for users’ locations selected at random within the cell range

User Data Rate vs. Users Density

Effective aggregate throughput as a function of the users density. Here the radius was changing inversely proportional to the users density.

Physical Layer

MAC Layer

MAC/Network QoS Mapping Layer

Intelligent Network QoS Validation Protocol Engine

Res. Alloc.

Network QoS

Intelligent Network QoS Validation Protocol

Interaction between the Physical/MAC layers and the Network QoS validation protocol

Intelligent Network QoS Validation Protocol (contd.)

Bandwidth Delay Jitter Loss

MAC/Network QoS mapping

MCS1

Application

MCS2

…MCSi

MCSi candidates

other

QoS classes

Traffic class Characteristics

Conventional Low delay, low date rate, sensitive

to delay variations, e.g. video conferencing

Streaming Less sensitive to delay, may

require high bandwidth, e.g. live coverage of sports

InteractiveBursty, variable bandwidth

requirement, moderate data loss, e.g. email, Telnet

BackgroundHigh tolerant to delay and data loss, high variable bandwidth, background file downloading

Intelligent Network QoS Validation Protocol (contd.)

At the MAC layer Network resources are allocated based

on the MAC QoS

At the network layer Network resources are validated based

on the network QoS requirements, which are traffic type dependent

Cross Layer Design Goal

Provide efficient methods of allocating network resources and applications

QoS support in all layers Dynamic protocol design Jointly optimize the network performance

Tradeoff Performance versus complexity and

scalability

Cross Layer Design (contd.)

Why? There exists direct coupling between

physical layer and upper protocol layers

Several upper layer protocols do not get advantage of the wireless medium information available within the physical and MAC layer.

Intelligent Network QoS Protocols

At the network layer

Intelligent network QoS validation protocol

Wireless QoS based routing protocol

Intelligent Network QoS Validation Protocol (contd.) Approach

Model the system as an objective function to be maximized based on network QoS parameters

Not to violate the MAC QoS constraints. Optimizing the objective function is an NP-

Complete problem Use heuristic techniques

Genetic algorithms Fuzzy logic Simulated annealing Etc.

Wireless QoS Based Routing Protocol

Features Low overhead control traffic On demand operation Optimal route computation Network configuration change Distributed operation

Communication

An optimum size of the protocol update control messages

The frequency of neighbor update messages

Modified version of the route discovery protocol implemented by most of the distance vector routing protocol

Wireless QoS Based Routing Protocol (contd.)

High Level Protocol Design QoS based routing protocol suite high

level design Optimization of resource usage

Advertise resource information to all neighboring nodes

Graceful network performance degradation Restrict the update message flooding to the

neighboring nodes only New neighboring information exchange policy

techniques using some heuristic algorithms

Adaptive routing protocol In case of a change in the network or node

resources, update messages will be triggered based on our neighboring information exchange policy

Granularity of the routing decision Source and destination-based routing

approach: the traffic between a given source and destination will be routed over the same route.

On the fly determination of feasible paths

High Level Protocol Design (contd.)

Performance objectives while computing QoS-based paths

Achieve better network throughput by achieving low route-request blocking probability while providing QoS based paths

Optimum routing overhead during the computation, communication, and routing information storage

Minimize the routing overhead caused by the rapid change of some of the network resources as well as the call setup frequency

High Level Protocol Design (contd.)

Route computation Use Kalman-filter or equivalent heuristic

techniques in order to determine the best path

Routing information storage Maintain a partial routing table

High Level Protocol Design (contd.)

Conclusion Improve the overall throughput of the

wireless network SINR AMC MCS Compute the best

threshold Use cross layer design technique Design a new protocol at the network layer

to assure better QoS based on the traffic type

Design a new QoS based routing protocol Less control message overhead Improve call blocking probability

Future Work

Simulation study using OPNET Performance evaluation of the

developed protocols (control message overhead, CBP, etc.)

Comparison study of the proposed QoS routing protocol with the existing protocols