Abstract
While the QoS and RB optimizations with instantaneous rate constraints feature well-established solutions, solving these problems with ergodic rates is demanding.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
These properties follow directly by replacing r k(t) with R k(t) within Sect. A.1 .
- 2.
Thus, the average SINR uses the channels second moments \(\boldsymbol {R}_k=\operatorname {E}[{\boldsymbol {h}}_k{\boldsymbol {h}}_k^{\operatorname {H}}]\) instead of \({\boldsymbol {h}}_k{\boldsymbol {h}}_k^{\operatorname {H}}\).
- 3.
Applying ZF beamforming for only a part of the receivers reduces the computational complexity. This is especially relevant if the number of antennas N and users K are large and N ≥ K.
- 4.
To simplify the notation, the noise variance is without loss of generality set to \(\sigma _k^2=1\).
- 5.
Instead of B k in (3.5), we can alternatively use , x↦A k(x) with
$$\displaystyle \begin{aligned} A_k(x) = \operatorname{E}[a_k(\zeta_k,x)] = \operatorname{E}[\operatorname{log}_{2}(\zeta_k+x)] = B_k(x)+\operatorname{E}[\operatorname{log}_{2}(\zeta_k)]. \end{aligned}$$While (3.5) leads to known closed-form expressions for Gaussian ξ k [see (3.6) and (3.7)], A k could be used if ζ k = (1 + ξ k)−2 were a positive quadratic form Gaussian random variables.
- 6.
- 7.
With , the random variable is \(\zeta _k^{-1}=|1+\xi _k|{ }^{2}=|w_k|{ }^{2}\), where .
- 8.
- 9.
The balancing target is ρ = 1 for the QoS optimization (3.1).
- 10.
The sum power constraint becomes \(\sum _{k=1}^K\|\boldsymbol {t}_k\|{ }_2^2=\sum _{k=1}^K\|\boldsymbol {\tau }_k\|{ }_2^2\leq P\), for example, [cf. (1.11)].
- 11.
This generalized ZF beamformer design has also been considered by [110], but for perfect CSI.
- 12.
See Sect. A.3 for the derivation of these bounds based on Jensen’s inequality and B k.
- 13.
According to Lapidoth and Moser [222, Lemma 10.1] (see also [223, Theorem 3]), the expected value for the logarithm of a non-central chi-square-distributed random variable \(y=\sum _{i=1}^N|x_i+m_i|{ }^2\) of degree 2N, with and i.i.d. , is given by
$$\displaystyle \begin{aligned} \operatorname{E}[\ln(y)] = \begin{cases} \ln(s^2) + \operatorname{E}_1\mspace{-2mu}(s^2) + \sum_{j=1}^{N-1}(-1)^j \left[ {\operatorname{e}}^{-s^2}(j-1)! -\frac{(N-1)!}{j(N-1-j)!} \right] \left( \frac{1}{s^2} \right)^j, &s^2>0\\ \psi(N), &s^2=0, \end{cases} \end{aligned}$$where \(s^2=\sum _{j=1}^N|m_i|{ }^2\), \(\operatorname {E}_1\mspace {-2mu}(\cdot )\) is the exponential integral [203, Chapter 5], and ψ(N) is the Digamma function [203, Section 6.3], e.g., \(\operatorname {E}[\ln (y)]=\ln (s^2)+\operatorname {E}_1\mspace {-2mu}(s^2)\) for N = 1 [224, Appendix B].
- 14.
Using \(R_k^{(\operatorname {B})}\) instead of \(R_{k,+}^{(\operatorname {B})}\) does not change the result as ρ k > 0 requires \(R_k^{(\operatorname {B})}(\boldsymbol {t})>0\).
- 15.
The set \(\mathcal {I}^{\operatorname {c}}\) stands for the complement subset to \(\mathcal {I}\), i.e., \(\mathcal {I}^{\operatorname {c}}=\{1,\ldots ,K\}\setminus \mathcal {I}\).
- 16.
The optimum of (3.22) could even become negative if \(R_{k,+}^{(\operatorname {B})}(\boldsymbol {t})\) was replaced by \(R_k^{(\operatorname {B})}(\boldsymbol {t})\).
- 17.
To simplify exposition, the target function is excluded from the interference function.
- 18.
The projection onto the non-negative orthant itself reads as \([z]^+=\max (0,z)\) for .
- 19.
This indexing is achieved via a reordering, such that \(\rho _1\rho -\mu _1^{(\operatorname {B})}\geq \ldots \geq \rho _K\rho -\mu _K^{(\operatorname {B})}\) for .
- 20.
- 21.
This precoder can be found numerically, as we have suggested in the previous section.
- 22.
These conditions ensure that none of the powers p k, k = 1, …, K increases unboundedly.
- 23.
Several other interference functions may be found for ergodic rates. The advantage of this version is that it only requires an evaluation of the involved \(\tilde {R}_k\) (3.4) for evaluating I k(p), k = 1, …, K.
- 24.
Similar to the case with instantaneous rates, (3.49) appears to be concave. Its bounds \(p_k/\tilde {R}_k^{(\operatorname {B})}(\boldsymbol {p})\) are concave. Moreover, \(p_k/\tilde {r}_k(\boldsymbol {p})\) is concave and \(\tilde {R}_k\) has similar properties as \(\tilde {r}_k\).
- 25.
It is impossible to further reduce the objective because \(\tilde {\boldsymbol {A}}\geq \mathbf {0}\) and \(\boldsymbol {p}^{\prime }\ngeq \boldsymbol {I}(\boldsymbol {p}^{\prime };\boldsymbol {\rho })\) if \(\exists i:p_i^{\prime }<p_i^\star \).
- 26.
Newton methods would also require the derivative of \(\tilde {R}_k\) with respect to p (n) in each iteration.
- 27.
For example, see [150] for the theory behind the sequential inner approximation strategy.
- 28.
We evaluate \(B_k^{-1}(y)\) with y > 0 numerically, using fixed point methods.
- 29.
The increase in c is approximately exponential due to the close relationship to the SINR.
- 30.
A rigorous mathematical proof for this observation is still missing.
- 31.
- 32.
We shortened notations by substituting α k := α k(1) and β k := β k(1).
- 33.
A feasible starting point t (0) is found by rescaling the solution of (3.16) based on UB2 (or LB1).
- 34.
- 35.
- 36.
Here, the matrix Y k(λ (n), μ) is independent of ρ (n+1) in contrast to the version in [143].
- 37.
All approximate interference constraints are active since α k(ρ) is strictly positive if ρ ≥ 0.
- 38.
Here, the effective noise variance is \(\nu _k^{(\operatorname {B})}=1\).
- 39.
The projected subgradient method does not result in a decreasing objective within each iteration.
- 40.
These problems are non-convex because the ergodic rates are neither convex nor concave functions in t and a convex reformulation of R k(t) ≥ ρρ k is missing.
- 41.
This BB problem formulation can be extended such that (p, t) resides in a combination of the generalized transmit power constraint set \(\mathcal {P}\) and the convex reformulations of instantaneous rate requirements if the transmitter is perfectly aware of some of the users’ channels [140].
- 42.
For the bound, we assume that all t i, i ≠ k are collinear with \(\bar {\boldsymbol {h}}_k\) and \(\|\boldsymbol {t}_i\|{ }_2^2=p^{\text{max}}\).
- 43.
In other words, the multiplicative factor is or .
References
P. Viswanath, D.N.C. Tse, Sum capacity of the vector Gaussian broadcast channel and uplink-downlink duality. IEEE Trans. Inf. Theory 49(8), 1912 (2003)
M. Schubert, H. Boche, QoS-based resource allocation and transceiver optimization. Found. Trends Commun. Inf. Theory 2(6), 383 (2005)
M. Rossi, A. Tulino, O. Simeone, A. Haimovich, Non-convex utility maximization in Gaussian MISO broadcast and interference channels, in Proceedings of the 36th International Conference on Acoustics, Speech, and Signal Processing ICASSP (2011), pp. 2960–2963
M. Kang, M.S. Alouini, Capacity of MIMO Rician channels. IEEE Trans. Wirel. Commun. 5(1), 112 (2006)
A. Gründinger, A. Barthelme, M. Joham, W. Utschick, Mean square error beamforming in SatCom: uplink-downlink duality with per-feed constraints, in Proceedings of the 11th International Symposium on Wireless Communication Systems ISWCS (2014), pp. 600–605
R. Hunger, M. Joham, A complete description of the QoS feasibility region in the vector broadcast channel. IEEE Trans. Signal Process. 57(7), 3870 (2010)
M. Schubert, H. Boche, Solution of the multiuser downlink beamforming problem with individual SINR constraints. IEEE Trans. Veh. Technol. 53(1), 18 (2004)
M. Bengtsson, B. Ottersten, Optimal downlink beamforming using semidefinite optimization, in Proceedings of the 37th Annual Allerton Conference on Communication, Control, and Computing (1999), pp. 987–996
W. Yu, T. Lan, Transmitter optimization for the multi-antenna downlink with per-antenna power constraints. IEEE Trans. Signal Process. 55(6), 2646 (2007)
A. Tölli, M. Codreanu, M. Juntti, Linear multiuser MIMO transceiver design with quality of service and per-antenna power constraints. IEEE Trans. Signal Process. 56(7), 3049 (2008)
M. Schubert, H. Boche, Iterative multiuser uplink and downlink beamforming under SINR constraints. IEEE Trans. Veh. Technol. 53(7), 2324 (2005)
M. Schubert, H. Boche, A generic approach to QoS-based transceiver optimization. IEEE Trans. Commun. 55(8), 1557 (2007)
H. Boche, M. Schubert, A superlinearly and globally convergent algorithm for power control and resource allocation with general interference functions. IEEE/ACM Trans. Netw. 16(2), 383 (2008)
N. Vučić, M. Schubert, Fixed point iteration for max-min SIR balancing with general interference functions, in Proceedings of the 36th International Conference on Acoustics, Speech, and Signal Processing ICASSP (2011), pp. 3456–3459
S. Boyd, L. Vandenberghe, Convex Optimization, 1st edn. (Cambridge University Press, New York, NY, 2004)
A. Gershman, N. Sidiropoulos, S. Shahbazpanahi, M. Bengtsson, B. Ottersten, Convex optimization-based beamforming. IEEE Signal Process. Mag. 27(3), 62 (2010)
A. Wiesel, Y. Eldar, S. Shamai, Zero-forcing precoding and generalized inverses. IEEE Trans. Signal Process. 56(9), 4409 (2008)
G. Dartmann, X. Gong, W. Afzal, G. Ascheid, On the duality of the max-min beamforming problem with per-antenna and per-antenna-array power constraints. IEEE Trans. Veh. Technol. 62(2), 606 (2013)
D. Cai, T. Quek, C.W. Tan, S. Low, Max-min weighted SINR in coordinated multicell MIMO downlink, in Proceedings of the 9th International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (2011), pp. 286–293
S. He, Y. Huang, L. Yang, A. Nallanathan, P. Liu, A multi-cell beamforming design by uplink-downlink max-min SINR duality. IEEE Trans. Wirel. Commun. 11(8), 2858 (2012)
B. Hassibi, B.M. Hochwald, How much training is needed in multiple-antenna wireless links?. IEEE Trans. Inf. Theory 49(4), 951 (2003)
A. Gründinger, M. Joham, W. Utschick, Design of beamforming in the satellite downlink with static and mobile users, in Proceedings of the 45th Asilomar Conference on Signals, Systems, and Computers, Pacific Grove (2011)
A. Gründinger, M. Joham, W. Utschick, Feasibility test and globally optimal beamformer design in the satellite downlink based on instantaneous and ergodic rates, in Proceedings of the 16th International ITG Workshop on Smart Antennas WSA, Dresden (2012), pp. 217–224
A. Gründinger, M. Joham, W. Utschick, Bounds on optimal power minimization and rate balancing in the satellite downlink, in Proceedings of the IEEE International Conference on Communications ICC (2012), pp. 3600–3605
A. Gründinger, A. Butabayeva, M. Joham, W. Utschick, Chance constrained and ergodic robust QoS power minimization in the satellite downlink, in Proceedings of the 46th Asilomar Conference on Signals, Systems, and Computers (2012), pp. 1147–1151
A. Gründinger, D. Leiner, M. Joham, C. Hellings, W. Utschick, Ergodic robust rate balancing for rank-one vector broadcast channels via sequential approximations, in Proceedings of the 38th International Conference on Acoustics, Speech, and Signal Processing ICASSP (2013), pp. 4744–4748
B. Marks, G. Wright, Technical note–a general inner approximation algorithm for nonconvex mathematical programs. Oper. Res. 26(4), 681 (1978)
H. Tuy, F. Al-Khayyal, P. Thach, Monotonic optimization: branch and cut methods, in Essays and Surveys in Global Optimization, Chap. 2 (Springer, New York, NY, 2005), pp. 39–78
J.P. González-Coma, A. Gründinger, M. Joham, L. Castedo Ribas, MSE balancing in the MIMO BC: unequal targets and probabilistic interference constraints. IEEE Trans. Signal Process. 65(12), 3293 (2017)
M. Abramowitz, I.A. Stegun, Handbook of Mathematical Functions, 1st edn. (Dover, New York, 1964)
E. Jorswieck, Lack of duality between SISO Gaussian MAC and BC with statistical CSIT. Electron. Lett. 42(25), 1466 (2006)
H. Boche, M. Schubert, A general duality theory for uplink and downlink beamforming, in IEEE 56th Vehicular Technology Conference, 2002. Proceedings. VTC 2002-Fall, vol. 1 (2002), pp. 87–91
J.P. González-Coma, M. Joham, P.M. Castro, L. Castedo, QoS constrained power minimization in the MISO broadcast channel with imperfect CSI. Signal Process. 131, 447 (2017)
A. Pastore, M. Joham, Mutual information bounds for MIMO channels under imperfect receiver CSI, in Proceedings of the 43rd Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, CA (2009), pp. 1456–1460
M. Joham, A. Gründinger, A. Pastore, J. Fonollosa, W. Utschick, Rate balancing in the vector BC with erroneous CSI at the receivers, in Proceedings of the 47th Annual Conference on Information Sciences and Systems, Baltimore, MD (2013)
A. Cuyt, V. Petersen, B. Verdonk, H. Waadeland, W. Jones, Confluent hypergeometric functions, in Handbook of Continued Fractions for Special Functions (Springer, Amsterdam, 2008), pp. 319–341
J. Lindblom, E. Larsson, E. Jorswieck, Parameterization of the MISO interference channel with transmit beamforming and partial channel state information, in Signals, Systems and Computers, 2008 42nd Asilomar Conference on (2008), pp. 1103–1107
A. Lapidoth, S. Moser, Capacity bounds via duality with applications to multiple-antenna systems on flat-fading channels. IEEE Trans. Inf. Theory 49(10), 2426 (2003)
S. Moser, Some expectations of a non-central Chi-square distribution with an even number of degrees of freedom, in 2007 IEEE Region 10 Conference (TENCON) (2007)
S. Moser, The fading number of memoryless multiple-input multiple-output fading channels. IEEE Trans. Inf. Theory 53(7), 2652 (2007)
C. Hellings, M. Joham, W. Utschick, Gradient-based rate balancing for MIMO broadcast channels with linear precoding, in Proceedings of the 15th International ITG Workshop on Smart Antennas WSA (2011)
C. Nuzman, Contraction approach to power control, with non-monotonic applications, in Proceedings of the IEEE Global Communications Conference GLOBECOM (2007), pp. 5283–5287
D.P. Bertsekas, Nonlinear Programming (Athena Scientific, Belmont, MA, 1999)
R. Hunger, D.A. Schmidt, M. Joham, W. Utschick, A general covariance-based optimization framework using orthogonal projections, in Proceedings of the 9th IEEE International Workshop on Signal Processing Advances in Wireless Communications SPAWC (2008), pp. 76–80
A. Bermon, R. Plemmons, Nonnegative Matrices in the Mathematical Sciences (Society for Industrial and Applied Mathematics, Philadelphia, 1994)
S. Stańczak, M. Wiczanowski, H. Boche, Fundamentals of Resource Allocation in Wireless Networks. Foundations in Signal Processing, Communications and Networking, vol. 3, 2nd edn. (Springer, Berlin/Heidelberg, 2008)
S. Stańczak, M. Kaliszan, N. Bambos, M. Wiczanowski, A characterization of max-min SIR-balanced power allocation with applications, in Proceedings of the IEEE International Symposium on Information Theory ISIT (2009), pp. 2747–2751
M. Grant, S. Boyd, Y. Ye, Disciplined convex programming, in Global Optimization: From Theory to Implementation, Nonconvex Optimization and Its Applications, ed. by L. Liberti, N. Maculan (Springer, New York, 2006), pp. 155–210
S.A. Vavasis, Complexity issues in global optimization: a survey, in Handbook of Global Optimization (Kluwer, Dordrecht, 1995), pp. 27–41
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Gründinger, A. (2020). Precoder Design for Ergodic Rates with Multiplicative Fading. In: Statistical Robust Beamforming for Broadcast Channels and Applications in Satellite Communication. Foundations in Signal Processing, Communications and Networking, vol 22. Springer, Cham. https://doi.org/10.1007/978-3-030-29578-3_3
Download citation
DOI: https://doi.org/10.1007/978-3-030-29578-3_3
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-29577-6
Online ISBN: 978-3-030-29578-3
eBook Packages: EngineeringEngineering (R0)