Authors: Zhanping Yin * Victor C. M. Leung Published: ACM/ MONET 2006 Presented by:

Third-Party Handshake Protocol for Efficient Peer Discoveryand Route Optimization in

IEEE 802.15.3 WPANs

Authors:Zhanping Yin * Victor C. M. Leung

Published: ACM/ MONET 2006

Presented by:Gautam S. Thakur

Presentation Topics

1. Definitions and Terminologies2. Knowing IEEE 802.15.33. Current Standards for Peer Discovery in IEEE

802.15.34. Issues and roadblocks

1. Peer Discovery and connectivity2. Ad hoc routing vs. MAC Layer forwarding

5. Proposed Peer Discovery protocol with forwarding route optimization (FRO)

6. Performance Evaluations1. Intra-piconet no connection probability2. Peer discovery delay analysis

7. Simulations and numerical results8. Conclusion and future goals

WPAN

• A WPAN (wireless personal area network) is a personal area network - a network for interconnecting devices centered around an individual person's workspace

• A very short range ~10 mtrs. E.g. Bluetooth

• Proposed operating frequencies are around 2.4 GHz in digital modes.

• The objective is to facilitate seamless operation among home or business devices and systems.

Ultra-Wideband

• A physical layer technology• For, short distance communication with

high data rate and short transmission range.

• Lower power requirement and pulsed data • UWB transmission over extremely wide

unlicensed radio spectrum 3.1 – 10.6 GHz.

• Over results in– Less interference – Wire-like performance for indoor wireless environment.

Presentation Topics

1. Definitions and Terminologies2. Knowing IEEE 802.15.33. Current Standards for Peer Discovery in IEEE

802.15.34. Issues and roadblocks

1. Peer Discovery and connectivity2. Ad hoc routing vs. MAC Layer forwarding

5. Proposed Peer Discovery protocol with forwarding route optimization (FRO)

6. Performance Evaluations1. Intra-piconet no connection probability2. Peer discovery delay analysis

7. Simulations and numerical results8. Conclusion and future goals

Overview of 802.15.3

• Why 802.15.3 ?

• The 802.15.3 Wireless Space

• 802.15.3 Overview and Components

Overview of 802.15.3[Why 802.15.3 ?]

• Motivated by the increasing demand of wireless communications with – Ubiquitous network connectivity – Low cost and low power consumption -> WPAN– High data rate(HDR)– Quality of Service(QoS) support– Comparison with other short to medium range wireless technologies

• Wireless LAN (WLAN) – High cost and power consumption, no hard QoS guarantee

• WPANs-Bluetooth (802.15.1) and ZigBee(802.15.4)– Data rate too low

• Applications of 802.15.3– Virtual wireless multimedia connectivity

• Video/audio distribution

– High speed data transfer

Overview of 802.15.3[ 802.15.3 = Wireless Multimedia ]

Overview of 802.15.3[ The 802 Wireless Space ]

IEEE 802.15.3 Overview • High date rate and low

power• Mainly works within a

piconet with dynamic DEV membership

• Ad hoc topology with centralized control by the PNC

• Connection oriented peer-to-peer communications

• Support for multimedia quality of service(QoS)

• Multiple power management modes

• Security

Formation of an 802.15.3 piconet

• The basic component is the DEV– One DEV is required to

assume the role of the piconet coordinator (PNC) of the piconet.

– The PNC provides the basic timing sync for the piconet with the beacon.

– Additionally, the PNC manages the quality of service (QoS) requirements, power save modes and access control to the piconet.

Formation of an 802.15.3 piconet (2)

• PNC supports ad hoc peer-to-peer connections

• PNC provides timing for synchronization of DEVs within the piconet, performs admission control, allocates network resources etc

Formation of a Piconet (3)

• All DEVs within radio coverage of the PNC can then associate with it to form a piconet.

• Then starts the peer discovery.

• Some DEV pairs in the piconet may be out of range of each other, and as a result, direct peer-to-peer connection is unavailable between them. This results in network layer discovery methods

Superframe format

• Timing and data transmissions in the piconet are based on the superframe

• The superframe has three parts– Beacon: Control information, Allocates CTA,

Synchronization– Contention Access Period (CAP): via CSMA/CA, file xfer– Channel Time Allocation Period (CTAP)

Presentation Topics

1. Definitions and Terminologies2. Knowing IEEE 802.15.33. Current Standards for Peer Discovery in IEEE

802.15.34. Issues and roadblocks

1. Peer Discovery and connectivity2. Ad hoc routing vs. MAC Layer forwarding

5. Proposed Peer Discovery protocol with forwarding route optimization (FRO)

6. Performance Evaluations1. Intra-piconet no connection probability2. Peer discovery delay analysis

7. Simulations and numerical results8. Conclusion and future goals

Peer discovery and connectivity issue

• An 802.15.3 piconet supports ad hoc communications between peer DEVs.

• Peer discovery is crucial to its operation.• The DEVs shall be able to obtain information

about the services and capabilities of other DEVs in the piconet at any time by information discovery commands.

• Peer information is needed before a source DEV can send any data to a destination DEV, or generate channel time requests (CTRq) to the PNC.

1

2

3 4

5

6 9

7 8

10 13

11 14

1512

PNC Src_DEV Dest_DEV

PNC Info. Request command

SIFSImm-ACK

PNC Information command

Imm-ACK

SIFS

SIFS

Probe Request Command

Imm-ACK

Probe Response Command

Imm-ACKSIFS

RIFS

SIFS

SIFS

SIFS

PNCDEV – 1 DEV - 2

Dev 2 is outside the range of Dev1

and vise-versa

1. NOT receive the Imm-Ack2. Cannot distinguish out-of-range

transmission and collision3. Perform backoff retransmission

for collision repeatedly

1. NOT receive Probe Request command

Existing Issues with IEEE 802.15.3• Peer discovery is crucial to

piconet operations. • Standard peer discovery is

unreliable and leads to substantial delays for unreachable DEV pairs

• Full piconet connectivity is not guaranteed with only direct peer-to-peer communications

• The standard 802.15.3 MAC does not take advantage of the unique ranging capabilities enabled by UWB

• Connections are in peer-to-peer manner without consider of possible route optimizations

• MAC modeling and performance evaluation

• Stream time scheduling methods not defined in the standard

• DEV_1 cannot communicate with DEV_4 in peer-to-peer manner

• For traffic between DEV_1 and DEV_3, is it better to forward via PNC than the direct connection?

• What is the optimal path and data rate between DEV_3 and DEV_5?

Using network layer routing

• Each hop request a CTA slot seperately.

• PNC treat each hop as independent traffic stream

• Effect: Failure in an intermediate hop breaks the connections, but PNC assumes that an independent stream is terminated. Keeps allocating CTAs for other hops untill it is eventually notified by all participating DEV

Using MAC layer forwarding

• With explicit MAC layer forwarding, PNC knows that these hops belongs to one connection.

• Better adjusting downstream and upstream CTAs.

• Effect: reroute the traffic is the intermediate node fails, and releasing all CTAs if source of destination terminates.

Some Observations• Peer discover is

essential to piconet operations and communication between DEVs

• Full piconet connectivity cannot be guaranteed

• Some DEV pairs may be out of range of each other

• Mac layer routing fails but is better then IP routing.

• Remember, 802.15.3 has centralized topology. So what ??

• Using PNC, any devices is two-hop count away only.

• Can use the central management capability to discover simpler and less costly MAC route

Proposed peer discovery protocol with forwarding route optimization

• The PNC can always act as a first hop to connect non-intersecting range DEVs

• However, a betterment can be done in the frame forwarding by choosing another DEV closer in distance to the source and destination to forward the frame.

• All routes are limited to two hops only.

Algorithm• If the Desti_DEV is reachable, sending PNC

Information Request and Response exchange is redundant.

• So, Src_DEV send Probe Request command to the Dest_DEV.

• If Dest_DEV receives the command, it returns an Imm-ACK after a SIFS and then the Probe Response command as in standard protocol operations.

• At the same time, the third party, i.e., the PNC, shall actively monitor the frame exchange.

• Upon receiving a Probe Request, the PNC checks the destination ID (Dest_ID) field in the MAC header.

Algorithm (2)• If the Dest_ID is not associated in the piconet, the PNC

send an Imm-ACK to the Src_DEV after SIFS, followed by a PNC Information command with an empty Information Element (IE) to notify the source that destination does not exist.

• Otherwise, instead of ignoring the Probe Request frame, the PNC waits for the Imm-ACK from the destination DEV.

• If no Imm-ACK arrives after a backoff inter-frame space (BIFS), which is the sum of a SIFS and a clear channel assessment detect time (CCADetectTime), the PNC realizes that the destination DEV cannot hear the source.

• The PNC then immediately send an Imm-ACK to the source, followed by a PNC Information command with the route information. (an optimized route information in sent)

SRC_DEV send Probe Request command to

DEST_DEV & PNC listens

DEST_DEV hears?

PNC also listen YES DEST_DEV responds

SIFS (Imm-ACK)

Probe Response command

DEST_DEV ID

Present?

No

Imm-ACK

SIFS

No

Yes, Timeout (BIFS)

Do Nothing, monitor if DEST_DEV crash.

Yes and DEST_DEV sent Imm-ACK

Send Imm-ACK & route information to source

Forward the data to the router node

Algorithm (3) [Route calculation]

• Most wireless networks today employ a multi-rate PHY (e.g., UWB in 802.15.3a) that supports a set of data rate dependent modulation/coding parameters.

• Due to the extremely low power consumption requirement of WPAN devices, the achievable data rate drops dramatically when the distance increases.

• The data rate can be modeled as a discrete function of transmission distance d between two DEVs:

Algorithm (3)

tyconnectivi no have DEVs theand

0 then ,an greater th is If ly.respective rates,

data maximum and minimum theare and and

; rate of rangeion transmiss theis ,0 where,

1

;)(

1

1

1

;1

Rate(d) Rd

SS

S R R

miRdR

SdRate

m

iim

ii

i

Algorithm (4)• Since the PNC can monitor all commands

exchanged during the CAP, it can learn the data rates between reachable DEV pairs and store them in an n x n rate matrix (RM), where

• n is the total number of DEVs within the piconet. DEV is assigned a unique DEV_ID in the piconet

. if and yet, availablenot

isn informatio rate or the range ofout are

and if 0 . and

between exchanged commands thefrom obtained

ninformatiocurrent on the based ratelink theis

ji RM

DEVDEV RMDEVDEV

RM

ij

j iijji

ij

Algorithm (5)

• All DEVs transmit with the maximum allowable power when sending data. Since the wireless links are symmetric in nature, clearly RMij = RMji.

• Based on the current rate information stored in RM, for an unreachable pair DEVi and DEVj, the PNC can determine the optimal (two-hop) route that has the minimum transmission time, i.e., the best route employs DEVk for MAC layer forwarding, where k minimizes:

Algorithm (6)

• Alleviates the traffic load on the PNC• Discovers a current optimal two-hop MAC layer

forwarding path given by existing rate information without introducing any extra overhead

• If the connection is broken, the PNC can immediately reroute the traffic to a current optimal path, or terminate the connection by de-allocating all corresponding CTAs.

]1,0[

)11

min(

nk

kjik RMRM

Discussion

• Original case: Causes retransmission

• 3PHP: data sent via 2-hop route if PNC_ID exists

Dest_ID

• Normal Transmission (Imm_Ack)

PNC_ID

• Sent Via 2–hop route (Imm_Ack)

Dest_ID

• Normal Transmission (Imm_Ack)

XXX FRAME LOST XXXX

• Causes retransmission

PNC Src_DEV Dest_DEV

PNC Info. Request command

BIFS

Imm-ACK

PNC Information command with Route Info.

Imm-ACK

SIFS

SIFS

Probe Request Command

RIFS

PNCDEV – 1 DEV - 2

3PHP-Node

Observations

• With 3PHP, peer discovery requires only one round of frame exchange.

• Fully utilized the broadcast nature, centralized control, ad-hoc communication, and efficient peer discovery.

• Save the futile back-off retransmission for unreachable destination.

• Guaranteed full piconet connectivity. (no network layer routing required)

• On demand routing• More then BIFS waiting. (include RIFS)

Conclusion

Performance Evaluations[Intra-piconet no connection probability]

• Intra-piconet no connection probability

• Common Overlap Area function COLA(R, r, x): Represent the intersection of two circles with radii R and r (r e R), respectively, which centers are separated by distance x

Performance Evaluations (2)[Intra-piconet no connection probability]

Performance Evaluations (3)[Intra-piconet no connection probability]

• The probability of no direct connection between two DEVs in a piconet is

Performance Evaluations (4)[Peer Discovery delay analysis]

•

# of Frame Transmission

• Expected • Contention Time

• Expected Packet Delay

Performance Evaluations (5)[Peer Discovery delay analysis]

• Expected routing failure probability

• Expected successful peer discovery delays are given by

Simulation and Numerical Results

• No connection probability as a function of coverage range ratio

Simulation and Numerical Results (2)

• Peer discovery delay vs. conditional collision probability

Simulation and Numerical Results (3)

• Piconet peer discovery time vs. coverage range ratio

Simulation and Numerical Results (4)

• Piconet peer discovery failure probability with standard method

Simulation and Numerical Results (5)

• Two-hop forwarding route optimization ratio vs. piconet radius

Simulation and Numerical Results (6)

• Expected data between directly unreachable pairs in piconets with 20 DEVs

Conclusion

• Underlying fact: peer-to-peer data delivery in 802.15.3 WPANs

• Existing MAC layer peer discovery methods cannot guarantee full connectivity between DEVs within a piconet through direct peer-to-peer connections if the piconet operates with a radius larger than half of the maximum transmission distance. (~ 41.3%)

• If Mac fails, use the expensive network layer routing.

Conclusion (2)

• Used the central control topology• Routing in 2-hop with single round of

control frame exchange• 3PHP achieves 25– 37% faster peer

discovery time over the standard MAC• Route optimization algorithm in the

PNC to provide the best MAC layer forwarding routes by self-learning the available rate information between DEVs.

References

• IEEE Standard 802.15.3, “Wireless medium access control (MAC) and physical layer (PHY) specifications for high rate wireless personal area networks (WPANs),”Sept. 2003.

• Z. Yin and V.C.M. Leung, “Third-Party Handshake Protocol for Efficient Peer Discovery in IEEE 802.15.3 WPANs,”in Proc. IEEE BroadNets2005, Boston, MA, Oct. 2005.

• Z. Yin and V.C.M. Leung, “Third-Party Handshake Protocol for Efficient Peer Discovery and Route Optimization in IEEE 802.15.3WPANs,”accepted for publication in ACM/KluwerJ. Mobile Networks and Applications, Nov. 2005.

• Z. Yin and V.C.M. Leung, “Connection Data Rate Optimization of IEEE 802.15.3 Scatternetswith Multi-rate Carriers,”IEEE ICC’06, Istanbul, Turkey, June 2006.

• Zhanping Yin and Victor C.M. Leung, “Introduction to IEEE 802.15.3 High Rate Wireless Personal Area Network (WPAN)”, Electrical and Computer Engineering University of British Columbia

Thank You