Top Banner
Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies
44

Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Dec 16, 2015

Download

Documents

Virgil Powell
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks

Ram RamanathanInternetwork Research Department

BBN Technologies

Page 2: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

MANET

Any multi-hop wireless network in which nodes relay packets for each other

Examples: Military Packet Radio NetworksSensor NetworksRooftop/Mesh Networks

Page 3: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Motivation

Despite decades of research, MANETs continue to lag behind wireline networks in terms of Latency Capacity Robustness

Need for Low-latency, High bandwidth wireless networks

Page 4: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Goals

Network with 1000+ Mobile Ad Hoc nodes Diameters (path-lengths) = 50-100 hops!! Transport capacity of 1 Gbps !! End-to-end latency less than 10ms Wireline robustness

Page 5: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Future prospects

Future military networks of sensors, robots, soldiers, ground, airborne vehicles

Hybrid wired/mobile-wireless civilian networks with large number of hops

…… … .

Page 6: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Where do we lack then..?

Page 7: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Reasons for severe under-utilization of performance potential

Hop Centric approach Unsuitable Physical Layer for

multi-hop/relay-based communications Failure to utilize broadcast nature of

MANETs

Page 8: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

A closer look..

Page 9: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Hop-centric approach

Processes are terminated and re-initiated at every hop

Large amount of processing, queuing and contention at each hop, for every packet

Each packet processed at 3 layers for header stripping

Page 10: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Bottleneck: Per packet overhead at each relay node

Page 11: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Subway train analogy

Its like getting off at each intermediate station en-route to one’s station

Going outside the station Waiting in line for fresh ticket Waiting for the next train Boarding it

Page 12: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Unsuitable Physical Layer

We still use Physical Layer suited for single-wireless-hop networks (WLAN/Cellular)

Current Physical Layer optimized for 2 primitives

Receiving Transmitting

Page 13: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

In MANETs

3 primitive operations required1. Relaying( Most common)2. Transmitting3. Receiving

CurrentlyRelay = Receive -> Store -> Process -> Queue -> Forward -> Contend -> Transmit

Page 14: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Failure to utilize Broadcast

We actually try to curb it by imposing wireline-like thinking

Most (traditional) routing protocols transmit to a single neighboring node

Broadcast can be used To increase signal quality End-to-end path capacity

Page 15: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Radical contributions..

Page 16: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Next generation MANET architecture

Three key features

1. Physical Layer optimized for multi-hop wireless networking

2. Access to medium for entire path (as opposed to single hop)

3. Cooperative transport of packets

Page 17: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

1. Physical Layer restructuring

Move “Routing” and “Forwarding” – functions to the physical layer!Routing:

To determine which set of nodes relay the packet from source to destination

Forwarding: To transport along this chosen path

Page 18: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

New Physical Layer

Has 3 primitive functionsRelayTransmitReceive

Switching at physical layer itself !

Page 19: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

2. Path-Centric hops

Atomic unit of operation = multiple hops

Medium Access Control is path-oriented

Packet does not have to re-contend at every hop

Page 20: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

3. Cooperative Transport

Harness unused resources to increase capacity of path

Concept of “Cooperative Diversity”Nodes simultaneously retransmit the same

packet on different frequencies/channels to be diversity combined at receivers

Page 21: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

How does this improve performance?

Reduced processing and elimination of re-contending at every hop will reduce latency

Cooperative transport increases capacity Path diversity increases path robustness

Page 22: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

ArchitectureNotional stack has 3 layers

1. Relay oriented Physical Layer(Relay PL)

2. Path Access Control(PAC)

3. Transport Layer

No Network Layer !!

Page 23: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Architecture

Page 24: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Important features

Paths are composed of “segments” A packet never leaves physical layer

throughout a segment PAC only invoked between segments Segment length: Interesting research

problem !

Page 25: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Lets Look in Detail…

Page 26: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

1. Relay-oriented Physical Layer

Based on a multi-frequency/multi-band system Full-duplex operation: Simultaneously

transmitting and receiving using multiple frequencies

Start transmitting while you are still receiving the rest of the packet

Transit Routing Table at Physical Layer for routing decisions

Page 27: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Relaying problems

Routing & Forwarding Essentially to decide at node X, for a

packet destined S -> D, whether toKeep packet (X=D)Discard it (X is not on path S -> D)Re-broadcast (Relay)

Page 28: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Mechanism

Extract certain information (destination/signal strength/..) from Front of the packet

Use it to decide whether to keep/drop/relay, while still receiving remaining packet

Shunt the incoming stream to transmit chain

Page 29: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Relay-Oriented Transceiver

Page 30: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Routing Decisions ?

Transit Control Table at a node X contains mappings from every source (S), destination (D) pair to one of keep/drop/relay

Proactive Link-State Routing run at Physical Layer

Routing updates and Neighbor discovery probes do not use the MAC layer

Page 31: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Link State Routing

Link State Updates (LSU) flooding when a link goes up or down

Flooding consists of a multihop network preamble followed by the actual LSU

Network Preamble “Captures” all nodes i.e. it gets them to ignore data transmission or reception and tune in to LSU

Page 32: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Routing features

“Capturing” of nodes ensures reliable broadcast of LSUs

As data rates increase what matters isPropagation time of updatesReliability of updates

Not how many control messages were sent!

Page 33: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Infrastructure for Relay PL

Hardware components well within scope of current technology

Routing logic & algorithms can be placed in Flash ROM (which are increasing in size & decreasing in cost)

Flexibility to use Software Radios – switching functionality can be in software

Page 34: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Before you ask me..

Page 35: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

There is no mention of any naming mechanism at the physical layer!!

Minor Implementation detail??

Page 36: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

2. Path Access Control (PAC)

Acquires the floor for multiple hops, namely a segment, within which packets are relayed at physical layer

Segment Access Request (SAR) = multi-hop RTS

Segment Access Clear (SAC)= multi-hop CTS

Page 37: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Sentinel

Page 38: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Important Issue

Setting up of frequencies of each node’s RX and TX to enable full-duplex operation

1. Select TX frequency and let RX “auto-tune”

2. (Less efficient) Always use SAR/SAC and decide a priory in half duplex mode

Any path can be full-duplexed using no more than 3 frequencies

Page 39: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Are you still awake ?

Just checking ;-)

Page 40: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

3. Cooperative Transport

Cooperative Diversity: Operates entirely at the Physical LayerNear simultaneous transmission of the same

information by multiple nodes that is coherently combined at the receiver

Gives much better SNR at receiver as essentially power of many nodes is added up

Page 41: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Cooperative Diversity Level of synchronization required for

decoding depends upon the receiver technology e.g. MIMO

MIMO or equivalent technology required to diversity-combine the simultaneous transmissions

frequency diversity: receiving multiple versions of the same signal, being transmitted at different carrier frequencies.

Page 42: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.
Page 43: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Future work

Developing h/w (Transceiver chipset) Determining optimal segment lengths Others…

Page 44: Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks Ram Ramanathan Internetwork Research Department BBN Technologies.

Thanks…

Ashish Sharma