The Road Towards xG A Swarm Perspective
Jan Rabaey
E. Alon, A. Arias, R. Brodersen, A. Niknejad,
B. Nikolic, V. Stojanovic, P. Wright
Berkeley Wireless Research Center
University of California @ Berkeley
Johannesberg Summit 2014
The Wireless Revolution to Continue
Nothing new here!
[Ericsson 2011]
15.9 ExaByte of Mobile Data by 2018
[Cisco 2014]
The Wireless Revolution to Continue
But in a far more diverse way
Tsensors Summit
Stanford, Oct 13
The Buzz around the
Swarm
The Changing Needs
[S. Parkvall, Ericsson 2013]
The Evolutionary Approach
[S. Parkvall, Ericsson 2013]
Let us sit back for a moment …
A Different Viewpoint
One Ring to Rule All … ?
A 21st Century Question in Wireless*
How would you construct a network if
wireless transceivers would
•
•
•
•
•
Be fully frequency agile
Be power scalable over wide range
Maintain multiple simultaneous channels
Not degrade in the presence of interference
Have substantial amounts of computational power
xGNet, aka the Fully Untethered Internet!
* Another “reverse time machine” – Hurray!
Not a day dream … really!
Configurable RF
(RF-FPGA)
• Frequency Agile MultiStandard Transmitters
• Multiple radios on a
single platform including
full-duplex
[Nikolic, Niknejad, Alon
BWRC, Nokia]
Not a day dream … really!
IMEC 60 GHz Transceiver [2013]
Integrated beamforming
using phased arrays
Not a day dream … really!
Multi-Beam/Frequency Hubs
[Courtesy: A. Niknejad]
xGNet
(aka the fully untethered internet)
• Continent - or globe-spanning wireless
network that obviates wired infrastructure
• Equally supporting all needs
– High data rates or small payloads
– Low latency or high reliability
– Good coverage or high energy efficiency
whenever needed!
• All at very small and scalable cost and
maximum efficiency
One possible way …
A more plausible way:
A Swarm perspective
The power is in
the numbers!
Exploiting density and locality
The power of
collaboration
Effciency
multi-hop [Gupta-Kumar00]
O(n(d-1)/2)
Example: mesh network
Energy
Spectral
Node density n
O(1)
O(1)
Node density n
Efficiency
Effciency
O(n1/2)
O(n)
O(n)
no-collaboration
Example: relay network
Node density n
But:
distributed MIMO [Tse07]
inherently self-limiting
[J. Rabaey, 2011]
A fresh look
Antenna Swarms
or Seas of Antennas (SOA)
Conjecture: Effective use of antenna arrays and
combinations thereof allow for virtual elimination of
interference and offer dramatic improvements in capacity
Example: Massive MIMO
• Substantial capacity, energy-efficiency, and flexibility
benefits available in array radio systems
[EU Mammoet Project]
17
A Scalable Approach:
FADER (eWallpaper)
power
printed
antenna
asic
Supports realization of arrays of arbitrary aperture, carrier
frequency, # of beams, etc.
[Courtesy: N. Narevsky, E. Alon]
18
Simple and cheap manufacturing and
assembly process
• Screen printed Ag antennas,
interconnect
19
• InkJet Flex copper power
lines
[Courtesy: A. Arias]
SOA’s not limited to local environment
100 Gbs/sec at 237 GHz
[Kalfass, KIT]
3 Gbs/sec E-band link
[www.e-band.com]
All of this possible in CMOS
or III-V
SOA’s not limited to local environment
xGNet: How?
Hierarchical mesh of xG-Hubs
– xG-Hubs could be managed or non-managed, fixed or
mobile, power-rich or poor
Basic principles:
– Antenna arrays create point-to-point links, avoiding
broadcast (== interference)
– Dynamic management of traffic (aka interference)
using all available parameters: frequency, space, time
– Distributed architecture that combines local
optimization with global objectives based on
economic incentives (aka connectivity brokerage)
Going One Step Further
• xGHubs more then access
points/routers
• Can provide substantial
computation and data storage
(caching!)
• Intelligent networking to stem
the data deluge!
Converging with the Swarm
Micron Hybrid Memory Cube
(HMC) [2014]
The High-Order Bits
• Antennas == Sensors/Actuators
• Swarms consisting of trillions of
deployed antennas offer new
opportunities on intelligent
networking
• The concept of almost “infinite
capacity” at high efficiency may
not be too far-fetched
[Warner Bros, 1978]