random access to wireless networks

34
Random Access to Wireless Networks Jean-Paul M.G. Linnartz Nat.Lab., Philips Research

Upload: cullen-hutchinson

Post on 04-Jan-2016

49 views

Category:

Documents


0 download

DESCRIPTION

Random Access to Wireless Networks. Jean-Paul M.G. Linnartz Nat.Lab., Philips Research. Random Access. Many terminals communicate to a single base station Fixed multiple access methods (TDMA, FDMA, CDMA) become inefficient when the traffic is bursty. Random Access works better for - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Random Access to Wireless Networks

Random Access to Wireless Networks

Jean-Paul M.G. LinnartzNat.Lab., Philips Research

Page 2: Random Access to Wireless Networks

Jean-Paul Linnartz 2

Radio Resource Management and Random Access

Random Access

• Many terminals communicate to a single base station

• Fixed multiple access methods (TDMA, FDMA, CDMA) become inefficient when the traffic is bursty.

• Random Access works better for

– many users, where ..

– each user only occasionally sends a message

Page 3: Random Access to Wireless Networks

Jean-Paul Linnartz 3

Radio Resource Management and Random Access

Suitable Protocols

• ALOHA

• Carrier Sense

• Inhibit Sense

• Collision Resolution– Stack Algorithm– Tree Algorithm

• Reservation methods– Reservation ALOHA– Packet Reservation Multiple Access

Page 4: Random Access to Wireless Networks

Jean-Paul Linnartz 4

Radio Resource Management and Random Access

ALOHA Protocol

• Any terminal is allowed to transmit without considering whether channel is idle or busy

• If packet is received correctly, the base station transmits an acknowledgement.

• If no acknowledgement is received by the mobile,1) it assumes the packet to be lost

2) it retransmits the packet after waiting a random time

Page 5: Random Access to Wireless Networks

Jean-Paul Linnartz 5

Radio Resource Management and Random Access

ALOHA Protocol• Any terminal is allowed to

transmit without considering whether channel is idle or busy

• If packet is received correctly, the base station transmits an acknowledgement.

• If no acknowledgement is received by the mobile,– 1) it assumes the packet to be

lost – 2) it retransmits the packet

after waiting a random time, usually with probability Pr in every slot.i t b

Page 6: Random Access to Wireless Networks

Jean-Paul Linnartz 6

Radio Resource Management and Random Access

ALOHA Protocol

• Unslotted ALOHA: transmission may start anytime• Slotted ALOHA: packets are transmitted in time slots

• Critical performance issue: "How to choose the retransmission parameter?"

– Too long: leads to excessive delay– Too short: stirs instability

Instability: Number of terminals in backlog grows without bounds

Page 7: Random Access to Wireless Networks

Jean-Paul Linnartz 7

Radio Resource Management and Random Access

Collision Resolution

• Adaptation of random retransmission

– global control of Pr (by base station)

– doubling the retransmission interval after every failure

• Dynamic frame length ALOHA• Stack algorithm• Tree algorithm

Page 8: Random Access to Wireless Networks

Jean-Paul Linnartz 8

Radio Resource Management and Random Access

Carrier Sense Multiple Access : CSMA– " Listen before talk "– No new packet transmission is initiated when the channel is busy – This reduces the number of collisions– Performance is very sensitive to delays in Carrier Sense

mechanism– CSMA is useful if channel sensing is much faster than packet

transmission time• satellite channel with long roundtrip delay: just use ALOHA

Hidden Terminal Problem:– mobile terminal may not be aware of a transmission by another

(remote) terminal. – Solution: Inhibit Sense Multiple Access (ISMA)– Decision Problem: how to distinguish noise and weak

transmission? – Solution: Inhibit Sense Multiple Access (ISMA)

Page 9: Random Access to Wireless Networks

Jean-Paul Linnartz 9

Radio Resource Management and Random Access

Inhibit Sense Multiple Access : ISMABusy Tone Multiple Access : BTMA

If busy, base station transmits a "busy" signal to inhibit all other mobile terminals from transmitting

Collisions still occur, because of– Signalling delay

• New packet transmissions can start during a delay in the broadcasting of the inhibit signal,

– Persistent terminals • After the termination of transmission, packets from persistent

terminals, awaiting the channel to become idle, can collide.

Page 10: Random Access to Wireless Networks

Jean-Paul Linnartz 10

Radio Resource Management and Random Access

Throughput - Offered Traffic (S-G) Relation

Page 11: Random Access to Wireless Networks

Jean-Paul Linnartz 11

Radio Resource Management and Random Access

Common Performance Analysis Assumptions

• All packets are of uniform duration, • unit of time = packet duration + guard time• Acknowledgements are never lost• Steady-state operation (stability)• Poisson distributed attempts

Steady-state operation:Random waiting times need to be long enough to ensure uncorrelated interference during the initial and successive transmission attempts. This is an approximation: dynamic retransmission control is needed in practice N.B. ALOHA without capture, with infinite population is always unstable

Page 12: Random Access to Wireless Networks

Jean-Paul Linnartz 12

Radio Resource Management and Random Access

WIRELESS RANDOM-ACCESS

Probability of successful reception depends on• Receiver capture performance

– Modulation method– Type of coding– Signal processing at the receiver (diversity, equalization, ...)

• Propagation:

– Distance from the central receiver, path loss, Shadowing – Channel fading and dispersion– Channel noise

• Traffic:

– Contending packet traffic (from same cell)– Interference from co-channel cells– Initial Access protocol: slotted ALOHA, Carrier Sense (CSMA) or Inhibit Sense Multiple

Access (ISMA)– Retransmission policy

Page 13: Random Access to Wireless Networks

Jean-Paul Linnartz 13

Radio Resource Management and Random Access

Capture effect and Near-far effect

• Not all packets in a collision are lost

• Nearby terminals have higher probability of success

• Unfairness

Page 14: Random Access to Wireless Networks

Jean-Paul Linnartz 14

Radio Resource Management and Random Access

Page 15: Random Access to Wireless Networks

Jean-Paul Linnartz 15

Radio Resource Management and Random Access

Cellular ALOHAScenario 1: Two separate

channels

Let’s assume we do time division:

Total traffic per cell is G packets per slot

Every available slot has 2G traffic because it carries traffic that arrives in two slot periods

System Throughput

S = 2G exp(-2G)

NB: Receiver is idle 50% of time

Scenario 2: Common Channel

Total traffic per cell is G packets per slot

Every available slot has 2G traffic because there are two cells

System Throughput

S = 2G exp(-2G)

or better if receivers can catch different packets

Page 16: Random Access to Wireless Networks

Jean-Paul Linnartz 16

Radio Resource Management and Random Access

Throughput versus offered traffic in wireless channel

ALOHA (orange), 1-persistent CSMA (green), non-persistent CSMA (blue)

Page 17: Random Access to Wireless Networks

Jean-Paul Linnartz 17

Radio Resource Management and Random Access

Cellular aspects of packet transmission

Circuit-switching:• Performance criterion is outage at cell boundary

Packet-switching:• Performance criterion is packet delay• Collisions from within the same cell, and• Interference from outside the cell• Optimum cell reuse” C=1

Page 18: Random Access to Wireless Networks

Jean-Paul Linnartz 18

Radio Resource Management and Random Access

DS-CDMA ALOHA Network

Under ideal signal separation conditions, DS-CDMA can enhance the capacity

Make a fair comparison!• Spreading by N in the same transmit bandwidth implies slot

that are N times longer. The arrival rate per slot is N times larger

Assumption for simple analysis: • All packets in a slot are successful iff the number of packets

in that slot does not exceed the speading gain.

Page 19: Random Access to Wireless Networks

Jean-Paul Linnartz 19

Radio Resource Management and Random Access

Intuition

• Compare the ALOHA system with an embarkation quay

• People arrive with Poisson arrival rate < 1 person per unit of time

• Boats of seat capacity N at regular intervals of duration N• Thus: total seat capacity is 1 person per unit of time• The boat sinks and the passengers drown if the number of

people exceeds N

6p

1p

Page 20: Random Access to Wireless Networks

Jean-Paul Linnartz 20

Radio Resource Management and Random Access

Intuition and AnalysisCase I: N = 1 (ALOHA without spreading)• Boats arrive very frequently• Probability of survival is exp{-G}

Case II: Large N (CDMA)• Fewer but larger boats arrive• Average waiting time is N times larger • Probability of survival is larger, because of the law of large numbers• Probability of success = Prob(n N) =

(-NG)n!

)(NG =)(

nN

1=nexpcaptP

Offered Traffic

Th

rou

gh

pu

t

Narrowband

“Ideal” CDMA

Page 21: Random Access to Wireless Networks

Jean-Paul Linnartz

Radio Resource Management and Random Access

6p

1p

Page 22: Random Access to Wireless Networks

Jean-Paul Linnartz

Radio Resource Management and Random Access

6p

1p

Page 23: Random Access to Wireless Networks

Jean-Paul Linnartz

Radio Resource Management and Random Access

6p

1p

Page 24: Random Access to Wireless Networks

Jean-Paul Linnartz

Radio Resource Management and Random Access

6p

1p

Page 25: Random Access to Wireless Networks

Jean-Paul Linnartz

Radio Resource Management and Random Access

6p

1p

Page 26: Random Access to Wireless Networks

Jean-Paul Linnartz

Radio Resource Management and Random Access

6p

1p

Page 27: Random Access to Wireless Networks

Jean-Paul Linnartz

Radio Resource Management and Random Access

6p

1p

Page 28: Random Access to Wireless Networks

Jean-Paul Linnartz

Radio Resource Management and Random Access

6p

1p

Page 29: Random Access to Wireless Networks

Jean-Paul Linnartz 29

Radio Resource Management and Random Access

Packet Delay versus Traffic

Page 30: Random Access to Wireless Networks

Jean-Paul Linnartz 30

Radio Resource Management and Random Access

Slow Frequency Hopping

• Narrowband (unspread) transmission of each packet• Receiver threshold remains unchanged• N parallel channels, each with rate 1/N• Traffic load per slot remains unchanged• Advantages:

– Frequency diversity: – fading on different carrier uncorrelated – capture probabilities independent – improved performance– Less Intersymbol Interference

• Disadvantages– longer delay than transmission at high rate

Page 31: Random Access to Wireless Networks

Jean-Paul Linnartz 31

Radio Resource Management and Random Access

Conclusions

• Wireless channel affects performance of random access scheme

– Packets lost due to fading, inter and intracell interference

– Signal capture enhances throughput and stability• Packet and circuit-switched data should be treated

differently • DS-CDMA in ISM bands does not necessarily help

Page 32: Random Access to Wireless Networks

Jean-Paul Linnartz 32

Radio Resource Management and Random Access

Bluetooth packet transmission• 1 Mbit/sec • Slots of 625 microsecond• Short packets• Frequency hopping

access code packet header payload

72 54 0 .. 2745 bits

Page 33: Random Access to Wireless Networks

Jean-Paul Linnartz 33

Radio Resource Management and Random Access

Bluetooth random access scheme• MULTI-SLOT PACKETSMULTI-SLOT PACKETS

f(k)

625 s

f(k+1) f(k+2) f(k+3) f(k+4)

f(k+3) f(k+4)f(k)

f(k)

f(k+5)

f(k+5)

f(k+5)

Page 34: Random Access to Wireless Networks

Jean-Paul Linnartz 34

Radio Resource Management and Random Access

Physical Link Definition

circuit switching symmetric, synchronous services slot reservation at fixed intervals

packet switching (a)symmetric, asynchronous services polling access scheme

SYNCHRONOUS CONNECTION-ORIENTED (SCO) LINK

ASYNCHRONOUS CONNECTION-LESS (ACL) LINK