A more Cognitive Approach to Wireless Access

CDT in Communica,ons A more Cognitive Approach to Wireless Access
Vaia Kalokidou; University of Bristol. Contact – [email protected]
Academic Supervisors: Dr. R.J. Piechocki & Dr. O. Johnson
Industrial Supervisors: Mr. Simon Fletcher & Ms. Patricia Wells (NEC)
Aim
§  Literature review on performance of wireless systems (focusing on MIMO), and transmission methods in MIMO systems
§  Robust and efficient extraction of wanted carrier in the presence of significant interference
Introduction
Transmission Methods in MIMO systems
Multiple-Input Multiple-Output (MIMO) systems
§  Numerous transmission methods exist, aiming at balancing interference mitigation with noise
enhancement
§  Linear receivers usually require a channel or frequency response estimate and suffer from a
certain amount of noise enhancement:
§  MIMO systems employ multi-element antenna arrays in Rx and Tx, obtaining high transmission
rates that grow linearly with the number of antennas in Rx and Tx
§  Applications: next generation cellular and WLAN products (WiMAX, 3GPP LTE)
- Zero-Forcing (ZF): employs a low complexity technique, utilizing all available DoF
- Minimum Mean Squared Error (MMSE): most popular suboptimal receiver, easily implemented
- Spatial Matched Filtering (SMF): can only be considered optimal for an orthogonal MIMO
channel
Interference
§  In cellular systems, it is created by BSs sharing
the same carrier frequency. In WLANs, it is created
by the presence of many APs in the same channel
§  Common methods for dealing with interference:
FDMA, TDMA, CDMA, SDMA
§  Successive Interference Cancellation (SIC) is an iterative receiver that decodes one data
stream at a time considering the rest of the data streams as interference. Usually, V-Blast and
D-Blast are based on a SIC receiver
§  Maximum Likelihood (ML) is optimal receiver,
BER for 4x2 2 user MIMO Rayleigh fast fading channel (BD algorithm − ZF Detection)
10
although it suffers from computational complexity
§  Multi-User MIMO:
Fig. 1 – Interference Channel [1]
Channel State Information (CSI)
0
- Block Diagonalization (BD): linear precoding
method, assuming perfect CSI at Rx and Tx
- Dirty Paper Coding (DPC): optimal precoding
method when full CSI is available
- Tomlinson-Harashima (TH): requires full
knowledge of the channel. High complexity
technique
Capacity in MIMO systems §  Capacity: maximal rate for which reliable
communication can be achieved.
§  In MIMO systems, it grows linearly with
number of antennas present at Rx and Tx
−1
10
BER
§  CSI refers to information known about the channel. Availability of CSI at Rx (CSIR) and Tx
(CSIT) has an effect on system performance
§  CSIR is easily obtained and includes all channel coefficients of the links from the Tx to Rx
§  CSIT is not easily obtained. Principles for acquiring CSIT: reciprocity and feedback
−2
10
−3
10
0
5
10
15
SNR (dB)
20
25
30
Fig. 5 – BER vs SNR for 4x4 MIMO with
BD precoding & ZF detection
Interference Mitigation Strategies
Fig. 2 - Capacity in a Rayleigh fading channel
for different number of antennas at Rx and Tx [2]
Fast Fading (MIMO) systems
§  Fast fading environment: for each transmission, channel is randomly chosen according to a
probability distribution and it is independent of input and noise.
§  In a Rayleigh iid fast fading MIMO system, where instantaneous value of CSIR and distribution
of CSIT are available, ergodic capacity can be determined as [3]:
where Nr, Nt: number of receive and transmit antennas respectively
Capacity CCDFs for a 4x4 Rayleigh i.i.d. MIMO system
Ergodic Capacity for a 4x4 Rayleigh i.i.d. MIMO system
1
25
0.9
0.8
0.7
15
Probability
Capacity(bits/Hz/sec)
20
10
0.6
0.5
increasing SNR
0.4
0.3
5
0.2
0.1
0
0
5
10
SNR(dB)
15
Fig. 3 – Ergodic Capacity (4x4 MIMO)
20
0
0
5
10
15
Capacity (bits/s/Hz)
20
25
30
Fig. 4 – Complementary CDFs (4x4 MIMO)
Slow Fading (MIMO) systems
§  Slow fading environment: for each transmission, channel remains constant. Channel fades
cannot be averaged out and it is challenging to ensure reliable communication
§  Performance is assessed by:
- Outage Probability: probability that selected transmission rate R is larger than maximal
achievable rate Rmax.
- Capacity with Outage: maximal rate for which outage probability is smaller than a
targeted outage probability po
§  For a Rayleigh iid slow fading MIMO system, outage probability is defined as [3]:
§  Many recent and innovative interference cancellation, avoidance and coordination strategies
exist. They are partially based on common interference mitigation techniques and transmission
methods
§  Examples: Interference Alignment (IA), CoMP-CSB scheme under user-selection algorithm,
Layered BD, Two stage MMSE (MMSE2), Beamforming with Joint Decoding (BF-JD)
Interference Alignment (IA)
§  IA is a cooperative interference management technique that exploits availability of multiple
signaling dimensions.
§  Aim: alignment of interference signals at each receiver
§  Transmitted signals are designed, in the cooperating transmitters, in such a way that interference
seen at receivers would occupy only a portion of the full signaling space. As shown in Fig. 6, three
interferers collapse to appear as two, therefore enabling interference-free decoding in a desired
signal subspace
§  Novelty: It attempts to align, rather than cancel or avoid, interference along dimensions different
from the dimensions of the actual wanted signal
§  Advantage: Every user can simultaneously
transmit at a data rate equal to half of their
interference-free channel capacity, allowing
network’s sum capacity to increase linearly with
no bound, with the size of the network
§  Challenge: Initial research on IA assumed perfect
CSI, which is difficult to acquire. Need for relaxed
assumptions
§  Consider a K user interference channel, where
every user i transmits a vector si, which is precoded
by the precoding matrix Fi. Then, the received signal
at user i is given by [1]:
where vi is the noise vector
§  Many algorithms on IA are being researched like
ZF-IA and Blind IA under different assumptions on
availability of CSI
Conclusions
§ 
§ 
§ 
§ 
Fig. 6 - IA concept [1]
Fig. 7 – ZF-IA compared to SMF [4]
In fading environments, where no CSIT is available, it is more challenging to define capacity and ensure reliable communication
Transmission strategies, such as ML and DPC are optimal, but only under full CSI assumption
Overall, acquisition of CSIT constitutes a big challenge in the design of interference mitigation algorithms
Novel algorithms are currently being researched, aiming at mitigating interference, even when CSIT is not known
Future Work
§  Further research on CSI acquisition, and interference mitigation strategies, mainly focusing on IA
§  Development of several models in MATLAB to investigate performances
§  Development of novel signal processing algorithms for interference mitigation in small cell
wireless networks
Acknowledgements I would like to thank my academic and industrial supervisors for their support and
guidance, and EPSRC and NEC for their financial support
References
[1] O. El Ayach, S.W. Peters, R.W. Heath, ‘The Practical Challenges of Interference Alignment’, ArXiv preprint arXiv: 1206.4755v1, June 2012
[2] D. Tse, P. Viswanath, Fundamentals of Wireless Communication, 1st ed., New York: Cambridge University Press, 2005
[3] T. Brown, E. De Carvalho, P. Kyritsi, Practical Guide to the MIMO Radio Channel, 1st ed., West Sussex: John Wiley & Sons Ltd, 2012
[4] C. Suh, M. Ho, D. Tse, ‘Downlink Interference Alignment’, IEEE Transactions on Communications, vol. 59, no. 9, pp. 2616-2626, Sept. 2011
This project is supported by the Engineering and Physical Sciences Research Council (EP/I028153/1) ; the University of Bristol and NEC.