This paper presents the use of the common-mode signal in digital-subscriber-line (DSL) transmission. The DSL binder's metallic sheath is used as a common reference to obtain reliable channels for the common-mode signals. The direct and crosstalk channels for the common-mode signal are obtained using multi-conductor transmission line theory, and the data-rate improvement by exploiting these extra dimensions is presented. The lines in the binder are coordinated to cancel differential-mode crosstalk and also to avoid interference caused by radiation from the common-mode within the binder. The high level of common-mode noise is countered by exploiting the correlation between the differential- and common-mode noises. Numerical results based on the derived models show data-rate improvements of up to 100% compared to using only differential-mode transmission.

This paper proposes a method that reduces transmit-power consumption of carrier-sense multiple-access (CSMA) networks by utilizing multiuser diversity and power control. Using this method, a terminal sends a packet at a slot if the terminal's signal-to-noise ratio (SNR) is above the threshold associated with the slot. Since the threshold value decreases as time advances, this method will make the user with the largest SNR access the shared channel; thus, a packet can be transmitted with less transmit power. The analysis under the infinite-user model shows that, as traffic load grows, the expected sum-power of the conventional CSMA network increases whereas that of the proposed method decreases.

In slow-fading uplink channels, carrier-sense multiple-access (CSMA) has been a popular random-access method, although it can have increased the throughput by using channel state information (CSI) acquired from a downlink training signal and electromagnetic reciprocity. To improve the throughput by using the CSI, this paper presents a random- access scheme that achieves multi-user diversity, called opportunistic p-persistent CSMA (OpCSMA). A user using OpCSMA sends a packet in a slot if the user's CSI exceeds the threshold assigned to the slot. The thresholds diminish with time to prioritize users having good CSI, while being designed to maintain the same persistency as the conventional CSMA, thus enabling backward compatibility. OpCSMA is mathematically shown to significantly increase the throughput compared with conventional CSMA, when the infinite-user model is used to model the variation of the number of users. As the expected number of users increases, the expected sum-rate of the OpCSMA approaches that using a centralized scheduler, which is the theoretical upper bound.

This paper characterizes the delay region that is quite useful for multi-user packet scheduling. In a quasi-static channel, the optimal delay region is defined as every set of total delay vectors achieved by some scheduling policies when there are no further packet arrivals. Each user's total delay is equivalent to the amount of time until that user's queue backlog is cleared. The optimal delay region shows the fundamental limit on each user's achievable queueing delay and describes the trade-off among the users in terms of queueing delay when different scheduling policies are applied. In addition, the concept of the delay region is shown to be extendable to each packet's average delay as well as total delay.

To increase the achievable sum rate of downlink MIMO (Multiple Input Multiple Output) systems, the cooperation among Base Stations (BS) is investigated in a slow fading channel. Three levels of BS coordination are considered: full coordination, partial coordination, and no coordination. This paper assumes that the downlink MIMO channel state information (CSI) is sent to the corresponding BS through a feedback link with some delay. In a fully coordinated system, the CSI at each BS needs to be collected at the central coordinator, and this process may result in significant additional delay. In this paper, the achievable throughput of each coordination level is evaluated and compared including the effects of the delay in CSI. A quasi-static fading is considered where the channel is static within one codeword, but it gradually changes over time based on the well-known Jake's model. Also, dirty paper coding (DPC) is applied, which is a transmission method shown to achieve the sum capacity of MIMO broadcast channels (BC). The results show that under certain delay conditions, the partially-coordinated system outperforms the fully-coordinated system. In addition, this paper presents a new partial-coordination method that provides higher throughput than the conventional partial-coordinated system.

This paper proposes an efficient algorithm that produces near-optimal discrete bit-loading for a Gaussian interference channel that models a DSL network. It first takes a continuous bit-loading initialization, and then rounds one user's bits in each tone independently, with rounding direction determined by evaluating a weighted sum rate objective function. After adjusting all the tones of a user, that user's power constraint is met by a power tightening procedure; then, the algorithm moves to the next user. Convergence of the proposed algorithm is proved. This algorithm can be made distributed by only allowing rounding DOWN the bits, and it can be also applied to Gaussian Multiple Access Channel (MAC) and Gaussian Broadcast Channel (BC) with slight modifications.

This paper addresses cross-layer resource allocation in quasi-static fading multiple access channels (MAC) with no channel state information at the transmitters (CSIT). Under the cross-layer approach without CSIT, the power and rate are determined based on an outage rate region as well as the current queue state information (QSI). For heterogeneous quality-of- service (QoS) requirements, an individual outage rate region results in better performance compared to the common outage rate region that is widely considered in the literature. This paper first reviews the recent notion of an individual outage rate region for the quasi-static fading MAC without CSIT. Individual outage rate regions are then used in two major types of scheduling policies: maximum weight matching scheduling (MWMS) and queue proportional scheduling (QPS). It is shown that MWMS and similar techniques based on weighted-sum-rate maximization require exponential complexity in the number of users because of the non-convex nature of the relevant optimization problems. By contrast, QPS can be very efficiently solved by using a successive feasibility check. Stochastic simulations show that compared to MWMS and the common outage rate region, significant throughput increase and delay reduction are possible by using the QPS scheduling policy with the individual outage rate region.

This paper introduces a multiple-input multiple-output channel model for the characterization of a binder of telephone lines. This model is based on multiconductor transmission line theory, and uses parameters that can be obtained from electromagnetic theory or measured data. The model generates frequency-dependent channel/binder transfer function matrices as a function of cable type, geometric line-spacing and twist-length parameters, and source--load configurations. The model allows the extraction of the magnitude and the phase of individual near end crosstalk, far end crosstalk, split-pair, and phantom transfer functions from the transfer function matrix of the binder. These individual crosstalk transfer functions are often found to be very sensitive to small imperfections in the binder. Examples of category 3 twisted pair American telephone lines and ldquoquadrdquo telephone cables are also presented.

In upstream vectored DSL transmission, the far-end crosstalk (FEXT) can be completely cancelled by using zero- forcing generalized decision-feedback equalizers (ZF-GDFE). When the spatially correlated alien crosstalk is present, the achievable data rates of DSL lines with ZF-GDFE depend on their decoding orders at each DMT tone. Given a weighted sum-rate maximization problem, the optimal orderings for all DMT tones can be found by the Lagrange dual decomposition method. However, the computational complexity of such approach grows with the factorial of the number of users, which makes the optimal search infeasible with a large number of vectored lines. This paper presents a modified greedy algorithm (MGA) that performs close to the optimal search of decoding orders. The complexity of MGA is only proportional to the cube of the number of users, which is the same as it of QR decomposition. With a significant reduction of complexity, MGA is a promising technique for practical DSL systems.

For multi-user orthogonal frequency division multiplexing (OFDM) systems, efficient optimal rate and power allocation algorithms are presented via geometric programming (GP), a special form of convex optimization problem for which very efficient interior point methods exist. Both multiple access channel (MAC) and broadcast channel (BC) are considered and the following two resource allocation problems are of main interest: weighted sum-rate maximization (WSRmax) and weighted sum-power minimization (WSPmin). Utilizing degradedness of BC on each tone, WSRmax and WSPmin in the BC can be all formulated as GP. By using the duality relation between MAC and BC, it is shown that the above resource allocation problems in the MAC can be converted into GP problems as well. This GP perspective of multi-user OFDM resource allocation problems provides numerical efficiency as well as strong scalability for any additional constraints of GP form.

In this paper, Opportunistic p-persistent Carrier Sense Multiple Access (OpCSMA) is proposed to incorporate multiuser diversity into p-persistent CSMA. In the OpCSMA, each mobile terminal (MT) is assumed to have the knowledge of its channel state information (CSI). At each idle time slot, an MT sends a packet if the corresponding channel gain is above the threshold, which is determined such that the probability of accessing the medium is maintained to be p for any idle slots. Using the infinite user model, it is proved that the OpCSMA provides significant increase in the expected capacity compared to p-persistent CSMA. In addition, the expected capacity of the OpCSMA is shown to be proportional to ln G at low SNR where G is the offered load. Simulation results corroborate the above findings in slow fading channels.

This paper proposes a power efficient p-persistent Carrier Sense Multiple Access (CSMA) employing multiuser diversity, called Opportunistic p-persistent CSMA (OpCSMA). At each idle time slot, an MT sends a packet if the corresponding channel gain is above the threshold which is determined such that the probability of accessing the medium is p for any idle slots. Also, the transmit power is controlled to maintain the constant signal-to-noise-ratio when the channel gain exceeds the cut-off fade depth. The analysis under the infinite user model shows that as traffic load grows, the transmit power consumption of the conventional p-persistent CSMA increases, whereas that of the OpCSMA decreases owing to multiuser diversity effect. Simulation results corroborate these findings in slow fading channels.

Queue Proportional Scheduling (QPS) assigns each user a data rate proportional to the number of packets (or bits) in that user's queue. This paper analyzes stability, delay and fairness properties of QPS in a Gaussian broadcast channel (BC). QPS is shown to achieve throughput optimality, and guarantee fairness as well as different priorities among users in terms of average queuing delay. One well known throughput optimal policy for broadcast channels is Maximum Weight Matching Scheduling (MWMS) that maximizes the inner product of the queue state vector and the achievable rate vector. Simulation results with Poisson packet arrivals and exponentially distributed packet lengths demonstrate that QPS provides a significant decrease in average queuing delay compared to MWMS in a Gaussian BC.

This paper proposes efficient rate and power allocation algorithms for OFDMA downlink systems where each tone is taken by at most one user. Weighted sum rate maximization (WSRmax) and weighted sum power minimization (WSPmin) problems are considered. Since these resource allocation problems are non-convex, complexity of finding the optimal solutions is prohibitively high. This paper employs the Lagrange dual decomposition method to efficiently solve both optimization problems. Because of their non-convex nature, there is no guarantee for the solution obtained by the dual decomposition method to be optimal. However, it is shown that with practical number of tones, the duality gap is virtually zero and the optimal solutions can be efficiently obtained

In a fading broadcast channel (BC), Queue Proportional Scheduling (QPS) is presented via geometric programming (GP). Given the current queue state, QPS allocates a data rate vector such that the expected rate vector averaged over all fading states is proportional to the current queue state vector as well as on the boundary of the ergodic BC capacity region. One well known throughput optimal policy for a fading BC is Maximum Weight Matching Scheduling (MWMS), which maximizes the inner product of the queue state vector and achievable rate vector. Simulation results for Poisson packet arrivals and exponentially distributed packet lengths demonstrate that QPS provides a significant decrease in average queuing delay compared to MWMS. In addition, QPS is shown to guarantee fairness among users in terms of average queuing delay.

For fading broadcast channels (BC), a throughput optimal scheduling policy called queue proportional scheduling (QPS) is presented via geometric programming (GP). QPS finds a data rate vector such that the expected rate vector over all fading states is proportional to the current queue state vector and is on the boundary of the ergodic capacity region of a fading BC. Utilizing the degradedness of BC for each fading state, QPS is formulated as a geometric program that can be solved with efficient algorithms. The GP formulation of QPS is also extended to orthogonal frequency-division multiplexing (OFDM) systems in a fading BC. The throughput optimality of QPS is proved, and it is shown that QPS can arbitrarily scale the ratio of each user's average queueing delay. Throughput, delay, and fairness properties of QPS are numerically evaluated in a fading BC and compared with other scheduling policies such as the well-known maximum weight matching scheduling (MWMS). Simulation results for Poisson packet arrivals and exponentially distributed packet lengths demonstrate that compared with MWMS, QPS provides a significant decrease in average queueing delay and has more desirable fairness properties.