Multiuser Multihop Networks

Motivation & Background

In the 1990�s the employment of multiple antennas at transmitter and receiver has been identified as the key enabler for high spectral efficiency in point-to-point communication, since it facilitates multiplexing of several data streams in space rather than only in time or frequency. As a consequence, the capacity of such a system scales linearly in the number of antennas at each terminal. In the meanwhile, MIMO is well understood in the context of point-to-point communication, and wireless communication research has shifted its focus towards the understanding of multi-user communication. A classical multi-user MIMO setting is the MIMO uplink, which corresponds, e.g., to the scenario that a set of mobile stations equipped with a single antenna each wishes to concurrently communicate to a base-station equipped with multiple antennas. In multi-user MIMO, an antenna array is not necessarily formed out of a set of collocated antennas, but can possibly be constructed out of a set of antennas that are distributed in space, e.g. antennas in mobile devices carried by different users. A second recent trend in wireless communication is the study of cooperative networks, i.e. networks containing non-selfish nodes. There are several strategies known, how cooperation between nodes can increase the aggregate network throughput, thus resulting in long-term benefits for all users in the network. Most of the strategies rely on nodes that � at a given time instance � assist the communication process without being interested in either sending or receiving any own data. A well established example of cooperative communication is so-called multi-hopping: the source signals traverse one or several relay stages (possibly consisting of multiple relay antennas each), before they reach the destination nodes. An obvious advantage of this kind of cooperation is an increase of the communication range.

The focus of this project is on multi-user MIMO multi-hop networks. Currently, only very little is known about such networks. One effect that is well understood is that a simple nonregenerative relaying strategy called �amplify & forward� comes along with severe problems in this setting � in particular when the length of the network grows large [1, 2]. According to the state of the art, it is not even clear, whether classical MIMO gains (such as linear capacity scaling in the number of source and destination antennas) can be sustained in long multi-hop networks. Against this background, the main contribution of this thesis shall be more elaborate signal processing and relaying strategies that better address the challenges coming along with this type of network and a better understanding of the fundamental limitations.

Contribution

The work to be pursued in this project focuses on multi-user MIMO communication in non-regenerative multihop networks (cf. Figure). In particular, we want to understand, how the dimensions of a MIMO multihop network, i.e. the number of hops and the number of relay antennas in each relay stage, impact the sum-capacity of the network, and thus the well known MIMO gains as they have been identified in point-to-point communication. In general, we assume that relay stages are composed out of a set of distributed antennas, which are not able to exchange any information about their corresponding receive signals. Likewise, source antennas are generally assumed to be distributed and non-cooperative. For the destination antennas both the collocated and the distributed case are investigated. The two cases are of fundamental difference in the sense that collocated destination antennas allow for decoding the code-words corresponding to the individual source nodes jointly based on the observations made at all destination antennas. In the distributed case, each destination antenna must decode the codeword of the corresponding source antenna without any knowledge about observations at other antennas, which renders interference-cancellation crucial. Due to the distributed nature of the source and relay antennas optimal signal processing in both the scenarios described above is non-trivial and not understood in the literature so far. Some exemplary questions, that are key for the understanding of the MIMO multihop problem are listed in the following:

Non-Coherent Amplify & Forward Relays, Collocated Destination Antennas: Whether or not it is possible to sustain linear capacity scaling in the number of source- and destination antennas in the limit of infinitely many hops by appropriately increasing the number of relay antennas per stage. Remark: If the number of relays per stage is of the order of the number of source and destination antennas, this is impossible according to [2].
Compress & Forward Relays, Collocated Destination Antennas: Can the more sophisticated relaying strategy in [3] (vector quantization and Wyner-Ziv compression) offer significant advantages over the amplify & forward strategy with respect to sum-capacity scaling of the network.
Coherent Amplify & Forward Relays, Distributed Destination Antennas: Is it possible to orthogonalize the multi-hop network through distributed beamforming at the relay nodes? What are necessary and sufficient conditions? Remark: In the special case of a two-hop network the answer is given in reference [4].

References

[1] S. Borade, L. Zheng, and R. Gallager, �Amplify and forward in wireless relay networks: Rate, diversity and network size,� IEEE Trans. Inform. Theory, vol. 53, no. 10, pp. 3302�3318, Oct. 2007.

[2] R. R. Mueller, �On the asymptotic eigenvalue distribution of concatenated vector-valued fading channels,� IEEE Trans. Inform. Theory, vol. 48, no. 7, pp. 2086�2091, July 2002.

[3] A. Sanderovich, S. Shamai (Shitz), Y. Steinberg, and M. Peleg, �Decentralized receiver in a MIMO system,� in Proc. IEEE Int. Symposium on Inf. Theory, Seattle, WA, July 2006, pp. 6�10.

[4] A. Wittneben and B. Rankov, �Distributed antenna systems and linear relaying for Gigabit MIMO wireless,� in IEEE Veh. Tech. Conf., Los Angeles, LA, Sept. 2004, pp. 3624�3630.

People	Joerg Wagner, Prof. Dr. A. Wittneben

Publications

Distributed Gain Matrix Optimization in Non-Regenerative MIMO Relay Networks
R. Rolny, J. Wagner, C. Esli, and A. Wittneben, Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, CA, USA, Nov. 2009.
On the Diversity-Multiplexing Tradeoff of Multiuser Ampfliy & Forward Multihop Networks
J. Wagner and A. Wittneben, Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, CA, Nov. 2009.
Distributed Gradient Based Gain Allocation for Coherent Multiuser AF Relaying Networks
C. Esli, J. Wagner, and A. Wittneben, IEEE International Conference on Communications (ICC 2009) , June 2009.
On Capacity Scaling of (Long) MIMO Amplify-and-Forward Multi-Hop Networks
J. Wagner and A. Wittneben, IEEE Asilomar Conference on Signals, Systems and Computers, Pacific Grove, CA, USA, Oct. 2008.
Multihop-Enabled Orthogonalization in Distributed MIMO Networks
J. Wagner, M. Kuhn, and A. Wittneben, IEEE Personal, Indoor and Mobile Radio Conference (PIMRC), Cannes, France, Sept. 2008, (invited paper).
Capacity Scaling of (Long) Non-Regenerative MIMO Multi-Hop Channels
J. Wagner and A. Wittneben, IEEE International Symposium on Information Theory, Toronto, ON, Canada, July 2008.
Cooperative Processing for the WLAN Uplink
M. Kuhn, J. Wagner, and A. Wittneben, IEEE Wireless Communications and Networking Conference, WCNC 2008, Las Vegas, USA, Mar. 2008.
On the Asymptotic Capacity of the Rayleigh Fading Amplify-and-Forward MIMO Relay Channel
J. Wagner, B. Rankov, and A. Wittneben, IEEE International Symposium on Information Theory, Nice, France, June 2007.