Logo
Home
Publications
Internal

New Protocols for Collaborative Transmission in Wireless Networks: Project Plan

Project manager

Erik G. Larsson
Associate Professor (docent)
Royal Institute of Technology (KTH)
School of Electrical Engineering, Communication Theory
Osquldas vag 10, Rm. B:408, SE-100 44 Stockholm, Sweden
Phone: +46-8-7908452 Fax: +46-8-7907260
Email: erik.larsson@ee.kth.se WWW: www.s3.kth.se/~elarsso

Background and Motivation:

Classically, radio link design and network protocol design have been treated as two separate disciplines. More recently, it has been understood that enormous gains in network throughput can be harvested by jointly designing access protocols and radio links. This has led to the paradigm of ``cross-layer design'', where several layers in the OSI model are optimized together. Such cross-layer design can bring gains of an order of magnitude in terms of capacity, data rate, and quality-of-service. One example is the use of smart packet scheduling algorithms that use the radio channel in an opportunistic fashion, for example by allocating the channel at a given time only to users with good fading conditions. Such opportunistic scheduling is already part of the 3G standard, embodied via HSDPA. Another example of cross-layer design, with even larger potential gains (but which lies slightly further into the future in terms of standards), is collaborative transmission. Basically, this is a way of letting terminals in a network cooperate with one another when they send packets to a given destination.

The idea of collaborative transmission has been around for some time, but only recently the industry has realized its enormous potential when used for coverage and data rate improvements in wireless networks. In its simplest form, collaboration takes the form of relaying; that is, letting a node A retransmit whatever she received from B, thereby increasing the chance that the data will reach their destination. Even such simple relaying has good potential and various issues related to it (e.g., the tradeoff between decode-and-forward and amplify-and-forward [1,2,3,4]) have been studied in some depth.

To understand the concept of collaboration, consider (for simplicity of the exposition) a wireless system with only two nodes, A and B, which wish to communicate packets to a destination node D, using a channel on which everyone can hear everyone else. Clearly A and B can transmit their packets to D without cooperation, simply by taking turns in using the channel (see Figure 1, top). Alternatively, they can cooperate by acting as relays for one another [1,2,3,4], as shown in Figure 1, middle. Here the available timeslots are divided into two subslots of half the length--one for the direct and one for the relayed transmission. This provides second order diversity even on a stationary channel (because two routes are available, namely $ A\to D$ and $ A\to B\to D$ respectively, for A's packets). Nevertheless this type of relaying is grossly suboptimal.

In the proposed project we will develop a new concept for user collaboration. The new concept is more sophisticated than simple relaying, yet not very complex. The idea is that nodes in the network not only relay information, but rather encode relayed data jointly with their own data using multiuser communication concepts. Our new concept is illustrated in Figure 1, bottom and described with more precision in the next section.

Technical Approach:

We explain our concept in the context of two collaborating nodes. The key idea is that when B, for example, acts as relay for A, it simultaneously transmits its own data packet and the packet for which it acts as relay, see Figure 1, bottom. The nodes then decode both messages using a form of multiuser detection.

Figure 1: Top: No collaboration. Middle: Classical collaboration. Bottom: New concept.
\begin{figure*} \centerline{ \resizebox{\textwidth}{7cm}{\epsfbox{larsson_CL2005_0236_fig1.eps}}} \end{figure*}

More precisely, transmission works as follows [5]. In the first slot, B transmits a superposition of its own packet $ B_1$, with the data received from A in the previous slot, $ A'_1$ (possibly encoding using a different channel code). Node B uses power $ 1-\gamma$ for its own packet, and power $ \gamma$ for $ A'_1$, where $ \gamma$, $ 0< \gamma\ll 1$, is a user parameter that can be optimized. Node then A decodes $ B_1$ by using a multiuser detector, consisting of a channel decoder and a soft MAP demodulator. In the second slot, A transmits its packet $ A_2$, and on top of this itsuperimposes $ B'_1$. The power level used for $ A_2$ is $ 1-\gamma$ and that used for $ B'_1$ is $ \gamma$. A joint decoder is then used at D to recover $ A_2$ and $ B_1$. The procedure then continues in slots 2 and 3 (for the transmission of $ A_1$), in slots 3 and 4 (for the transmission of $ B_2$) and so forth.

The new concept substantially outperforms classical cooperation, as illustrated in Figure 2 for some different spectral efficiencies. From the figure it is clear that both classical cooperation and the new scheme can extract second order diversity, however, the new scheme also provides an additional coding gain. This gain grows with the spectral efficiency, which shows that gains from collaboration can be significant even at high rates (in contrast to relaying-based collaboration). Also, notably and perhaps somewhat surprisingly, it turns out that the encoding/decoding complexity of the new scheme is precisely the same as that of classical cooperation. This is so because both schemes use signal constellations of the same effective size and codewords of the same length, so demodulator and decoder complexity will be the same. The performance gain over classical relaying-based collaboration is therefore obtained ``for free''.

Research Directions for the Project:

Our new concept is very promising because it circumvents the suboptimality associated with channel orthogonalization, which is the main problem of existing pure relaying-based collaboration approaches [1,2,3,4]. In the research project we will take our new concept further in the following directions:
  1. Explore whether, and how the concept can be extended to the the multiterminal collaboration scenario. That is, several users A, B, C, ... wish to communicate packets to a common destination. Analyze how this protocol behaves, what the tradeoffs are, and whether the statement of equal-complexity (compared to pure ``relaying'') would translate to the multiterminal scenario. Investigate scaling aspects, and how the gain over classical relaying grows with the number of involved nodes.
  2. Design and analyze the optimal receiver for the new scheme, especially when more than two nodes are involved. The receiver used in Figure 2 was suboptimal in the sense that only one packet at a time is decoded. However, due to the nested structure imposed by the protocol in Figure 1, bottom, received data at a given time will contain information on all previously transmitted packets. Joint decoding of several packest can therefore potentially improve performance even further, at no or little extra complexity. One important issue to investigate here is how to avoid that error propagation degrades performance.
  3. To explore the connections between to so-called ``dirty-paper coding'', in particular investigate whether techniques for dirty-paper coding can further improve our scheme.

The proposed work is basic research, but as illustrated in [2] user collaboration has enormous potential and may be an option for coverage enhancement in future cellular systems.

Figure 2: Packet error rate as a function of the signal-to-noise ratio (SNR) for spectral efficiencies $ 0.66$ and $ 1.33$ bps/Hz. The packet length was 1000 bits. More detail is available in [5].
\begin{figure}\centerline{ \resizebox{8cm}{7cm}{\epsfbox{larsson_CL2005_0236_fig5.eps}}}\end{figure}

References:

1
R. U. Nabar, F. W. Kneubuhler, and H. Bölcskei, ``Performance limits of amplify-and-forward based fading relay channels,'' in Proc. of IEEE ICASSP, June 2004.

2
A. Sendonaris, E. Erkip, and B. Aazhang, ``User cooperation diversity, part I: System description,'' IEEE Trans. Commun., vol. 51, pp. 1927-1938, Nov 2003.

3
J. N. Laneman, D. N. Tse, and G. W. Wornell, ``Cooperative diversity in wireless networks: Efficient protocol and outage behavior,'' IEEE Trans. Inform. Theory, vol. 50, pp. 3062-3080, Dec. 2004.

4
T. E. Hunter and A. Nosratinia, ``Cooperation Diversity Through Coding,'' Proc. of IEEE ISIT, Lausanne, Switzerland, 2002.

5
E. G. Larsson and B. Vojcic, ``Cooperative transmit diversity based on superposition modulation,'' IEEE Communications Letters, vol. 9, pp. 778-780, Sept. 2005.

Last updated: 2006-02-06 by Niklas Olsson