computer networks group universität paderborn tandem project meeting protocols, oversimplification,...
TRANSCRIPT
Computer Networks GroupUniversität Paderborn
TANDEM project meeting Protocols, oversimplification, and cooperationor: Putting wireless back into WSNs
Holger Karl
2
Overview
Protocols in WSN and oversimplified models
Cooperation diversity – motivation & approach
Cooperation diversity in wireless sensor networks Relationship to 802.15.4 Relationship to TANDEM candidate routing scheme
3
Routing and oversimplified models
To a computer scientist, a network is a graph Well-behaved, stable Resulting model: Unit disc graph
To an electrical engineer, a network is a collection of channels
Wireless channels: fickle, changing, hard to predict
4
Protocols in unit disk graphs
Under unit disk graph assumptions (and similar): Protocol design is easy…
Neighbors stay neighbors Error control is easy Dynamics in the network are slow
… but futile Networks do not behave like this Typical problem: Short- and long-term fading
! Variations in network on short time-scales than
! Need to design protocols accordingly
But not all is hopeless!
5
Cooperation diversity – motivation
Fading is a source of serious problems But also an opportunity
Transmission A ! C fails (fading) Transmission A ! B works (wireless multicast advantage!) Transmission B ! C would work – but should B cooperate?
A possible gains from cooperation in presence of channel diversity ! cooperation diversity
A
B
C
6
Store-and-forward vs. cooperative relaying
Store-and-forward:
TDMA example:
User cooperative diversity:
TDMA example:
AB
D
A
B
D
A
B
A
B
A
B
Time
A
B
Phase1 Phase2 Time
A
B
Phase1 Phase2
Each relaying scheme employs: Two phases, two types of channels Phases: Reception/retransmission Channels: Inter-user/uplink
7
Cooperation diversity – known properties
Cooperation diversity substantially improves outage probability
Specifically advantageous if “inter-partner” channel is good Has many variations
How does “partner” behave when an incorrect packet arrives? How to forward a packet? Completely? Or partial redundancy?
8
Schemes: Regeneration vs. non-regeneration
Objectives:1. Do not transmit redundancy if user does not need help
Requires a priori knowledge of uplink channel state
2. Do not relay erroneous data in phase 2 Requires regeneration and error-checking partner’s data
Non-regenerative schemes: e.g. Amplify-and-Forward, saves energy but relays possibly
erroneous data ) Useful for “good” inter-user channels
Regenerative schemes: Enable error-checking Decode-and-Forward: Repeat partner’s data
Correct reception in phase 1: Partner’s data is repeated in phase 2
Wrong reception in phase 1: Own data is transmittedA B A B
Phase 1 Phase 2
B!A, A‘s decoding fails:
9
Coded cooperative schemes, Cooperation level
Regeneration further enables: More efficient coding Instead of repeating partner’s data:
Relay only parts (redundancy) in phase 2 Use efficient FEC and/or space-time codes
Many approaches for these coded cooperative schemes Whitepaper provides detailed discussion
With coded cooperation redundancy can be adjusted: Cooperation level controls proportion of time used for diversity Example: Long phase 1 (less red.) if partner’s uplink channel good
) Channel-aware div/mux optimization: If we can estimate uplink channel state we can adjust phase
lengths, i.e. user cooperative diversity accordingly
A B A B
Phase 1 Phase 2
1/2: A B A B
Phase 1 Phase 2
3/4:
10
Cooperation diversity and wireless sensor networks?
Applicable to wireless sensor networks movement exists Nodes themselves Or object in the environment!
Concrete problems High density of WSNs – how to handle possibly many partners? How to balance energy constraints and willingness to cooperate
Similar to multi-hop forwarding, but possibly with different trade-offs if nodes are allowed to forward own data along with cooperation data
How to handle energy consumption, tradeoffs vs. improved error probabilities?
11
Cooperation in TANDEM: MAC protocol
TANDEM assumes a commodity-off-the-shelf MAC/link layer to improve compatibility – IEEE 802.15.4
Possible to Integrate cooperation into existing 802.15.4? Extend 802.15.4 by cooperation? With minimal impact on compatibility?
Integrate partner selection into a wakeup mechanism?
Consequences for energy efficiency, implementation complexity, … ?
12
Cooperation in TANDEM: Routing protocol
TANDEM proposed routing mechanism: Use routes with sparse neighborhood
To circumvent false wakeup/idle listening
This trade offs against the smaller number of possible partners
What are the trade-offs? Possible to characterize when to avoid dense regions?
When the channel conditions in the sparse region are approximately AWGN?
Integrate such knowledge into cooperation scheme?
13
Conclusion
Cooperation diversity is a promising mechanisms to handle wireless channel imperfections in networks
Specifically, lowers error probability
Application to wireless sensor networks little researched
Promising research questions in TANDEM Integration into MAC, routing – geared towards WSN
14
The coded cooperation approach
Additional coding of data across spatial separated users
A simple cooperative coding scheme:1. Select partner to cooperate
and cooperation level 2. Both partner transmit first
subframe using code rate R1=R/
3. Each node decodes data; if no errors occur cooperation is possible
4. With cooperation: Each node transmits redundancy for its partner to ensure that full code-rate R is reached Without cooperation: Node
transmits redundancy for its own data
Gain: If N user cooperate scheme achieves full diversity order N, at high SNR this means up to 270% spectral efficiency
Similar gains are provided by comparable approaches, e.g. decode-and-forward, space-time cooperation
Data B
Data A
1.2.
3.
3.
Red. B
Red. A
4.
4.
2.