[Show abstract][Hide abstract] ABSTRACT: Binary power control has been shown to be optimal in a variety of network conditions. In the context of sumrate maximisation in interference networks, our main result states that the optimal power policy for a link is binary if it only interferes with one other link in the network. Examples of networks which have at least one link with this attribute include the two-link interference channel (IC), the one-sided Wyner network and the many-to-one interference channel. Previous works have fully characterised the optimal power scheme for the two-link IC and they agree with our results. For the one-sided Wyner network, our main result leads to a problem which is tractable and can be solved using dynamic programming methods.
Proceedings of the Global Communications Conference, 2010. GLOBECOM 2010, 6-10 December 2010, Miami, Florida, USA; 01/2010
[Show abstract][Hide abstract] ABSTRACT: We consider the power optimization problem of maximizing the sum rate of a symmetric network of interfering links in Gaussian noise. All transmitters have an average transmit power constraint, the same for all transmitters. We solve this nonconvex problem by indentifying some underlying convex structure. In particular, we characterize the maximum sum rate of the network, and show that there are essentially two possible states at the optimal solution depending on the cross-gain between the links, and/or the average power constraint: the first is a wideband (WB) state , in which all links interfere with each other, and the second is a frequency division multiplexing (FDM) state, in which all links operate in orthogonal frequency bands. The FDM state is optimal if the cross-gain between the links is above 1= p 2. If the cross-gain is below 1= p 2, then FDM is still optimal provided the average power of the links is sufficiently high. Assuming that < 1= p 2, we can say that the WB state occurs when the average power level is low (relative to the noise and the cross-gain factor between the links), but as we increase the average power level from low to high, there is a smooth transition from the WB state to the FDM state: For intermediate average power levels, the optimal configuration is a mixture, with some fraction of the bandwidth in the WB state, and the other fraction in the FDM state. This work has applications to DSL, as well as to wireless networks.
IEEE Transactions on Information Theory 01/2010; 56:4471-4487. DOI:10.1109/ICC.2009.5199170 · 2.33 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: In a Wyner model each link only interferes with links adjacent to it. We consider the problem of finding the optimal power allocation which maximizes the sum-rate of such a network. Each link has a maximum power constraint and the power allocation is assumed to be time and frequency flat. In the case of 3- and 4-link Wyner models, we show that the optimal power schemes are in fact binary, i.e. a link is either switched off or turned on at full power. The problem is then extended to larger-sized Wyner models by limiting to optimal binary power schemes. Interesting phase transitions are observed as the interference cross-gain, ε, traverses various thresholds.
Proceedings of IEEE International Conference on Communications, ICC 2010, Cape Town, South Africa, 23-27 May 2010; 01/2010
[Show abstract][Hide abstract] ABSTRACT: We consider the power optimization problem of maximizing the sum rate of a symmetric network of interfering links in Gaussian noise. All transmitters have an average transmit power constraint, the same for all transmitters. We solve this nonconvex problem by indentifying some underlying convex structure. In particular, we characterize the maximum sum rate of the network, and show that there are essentially two possible states at the optimal solution depending on the cross-gain (√ε) between the links, and/or the average power constraint: the first is a wideband (WB) state , in which all links interfere with each other, and the second is a frequency division multiplexing (FDM) state, in which all links operate in orthogonal frequency bands. The FDM state is optimal if the cross-gain between the links is above 1/√2. If ε < 1/2, then FDM is still optimal provided the average power of the links is sufficiently high. With ε < 1/2, the WB state occurs when the average power level is low (relative to the noise and the cross-gain factor), but as we increase the average power level from low to high, there is a smooth transition from the WB state to the FDM state: For intermediate average power levels, the optimal configuration is a mixture, with some fraction of the bandwidth in the WB state, and the other fraction in the FDM state. This work has applications to DSL, as well as to wireless networks.
Proceedings of IEEE International Conference on Communications, ICC 2009, Dresden, Germany, 14-18 June 2009; 02/2009
[Show abstract][Hide abstract] ABSTRACT: We consider the power control problem of maximizing the sum rate of a symmetric network of interfering links in Gaussian noise. We consider a static network: there is no time-varying fading and the power allocation is also mandated to be time and frequency flat. All transmitters have a maximum allowable average transmit power, the same for all transmitters. We solve this nonconvex problem by identifying some underlying convex structure, and show that the solution is either one link blasting at full power, or all links blasting at full power. We provide a characterization of the solution in terms of the level of cross-gain between the interfering links. There is a phase transition between these two states, as the cross-gain traverses a threshold.
[Show abstract][Hide abstract] ABSTRACT: In this paper, we consider the problem of maximising the sum rate of two interfering links in Gaussian noise. We assume no fading and that interference is treated as worst-case Gaussian noise, and that the optimal power allocation must be time-invariant. We show that either both links must operate at maximum power, or one link operates at maximum power and the other link is switched off. The switching point between one power scheme to another depends on the value of gamma which is the SNR of the link, as well as the channel gains at all links.
Communications Theory Workshop, 2008. AusCTW 2008. Australian; 01/2008
[Show abstract][Hide abstract] ABSTRACT: Motion estimation (ME) is an important part of H.264 and other video compression standards, which eliminates temporal redundancies at the expense of higher computational complexity. Many adaptive methods have been proposed to reduce the amount of computation involved in ME. In this paper, two improvements are added to the technique using directional adaptive search window (DASW) method: directional search process and early termination. DASW alone can achieve a savings in number of search point computations of between 42% to 60%. By introducing early termination(ET) into motion estimation using DASW, the number of computations could be reduced even further since the search process is terminated immediately if the block is considered as a stationary block. However, DASW-ET incurs a slight degradation of less than 1 dB, in PSNR performance.
Intelligent and Advanced Systems, 2007. ICIAS 2007. International Conference on; 12/2007
[Show abstract][Hide abstract] ABSTRACT: Motion estimation (ME) is the exploitation of similarities between adjacent frames in a video sequences, and is an essential part of the H.264 and other video compression standards. However, it introduces an increase of computation complexity resulting in longer execution time. In this paper, the search range for block-matching ME is made adaptive, based on the motion vectors present in preceding frames. The proposed algorithm determines the amount of motion in past frames and classifies the video sequence as having low, medium or high motion. This classification and the direction of the motion vectors, are used to dynamically adjust the search window size and general orientation. Simulation results show that execution time could be reduced by 42%-60% with a small degradation in PSNR value, to that of non-adaptive ME
TENCON 2006. 2006 IEEE Region 10 Conference; 12/2006
[Show abstract][Hide abstract] ABSTRACT: The H.264/AVC video compression standard incorporates new features to enhance video coding. It supports a variety of block sizes and shapes for motion estimation (ME), and gives more flexibility than previous standards. In this paper, we propose a method of determining the block mode to be used for a particular video sequence to give the optimum performance for the encoder. To reduce the number of computations involved, the block mode decision phase is done on the first few frames only and fixed for subsequent frames. The decision is based on motion vectors (MVs) found from previous frames. From our results, the block computation time can be reduced as much as 47.72% for a video sequence with low motion activity. On average, most of the test video sequences suffer degradation in PSNR value of less than 0.5 dB