Hon Fah Chong

Mobile Data Offloading Through A Third-Party WiFi Access Point: An Operator's Perspective

WiFi offloading is regarded as one of the most promising techniques to deal with the explosive data increase in the existing cellular network due to its high data transmission rate, and low requirement on devices. In this paper, we investigate the mobile data offloading problem through a third-party WiFi access point (AP) for a cellular mobile system. From the cellular operators perspective, by assuming the data offloaded through the third-party WiFi AP is charged based on usage, we formulate mobile data offloading problem as a utility maximization problem, which is an integer programming problem. A centralized mobile data offloading scheme is then obtained by considering its relaxation problem. It is strictly proved that the proposed centralized data offloading scheme is near-optimal when the number of users is large. Then, to relieve the computing burden on the cellular base station, we propose a threshold-based distributed data offloading scheme. We show that the proposed distributed scheme performs very well once the threshold is properly chosen.

New coding strategies for the relay channel

This paper studies coding strategies for a three-node relay channel. We first review the basic coding strategies of Cover and El Gamal for the relay channel. Next, two new coding strategies superimposing cooperation and facilitation are developed. One of the coding strategies is shown to include the generalized strategy of Cover and El Gamal. For certain parameters of the Gaussian relay channel, the two new strategies give higher achievable rates than the generalized strategy of Cover and El Gamal

Ergodic Sum-Rate Maximization for Fading Cognitive Multiple-Access Channels Without Successive Interference Cancelation

In this paper, the ergodic sum rate of a fading cognitive multiple-access channel (C-MAC) is studied, where a secondary network (SN) with multiple secondary users (SUs) transmitting to a secondary base station shares the spectrum band with a primary user (PU). An interference power constraint (IPC) is imposed on the SN to protect the PU. Under such a constraint and the individual transmit power constraint (TPC) imposed on each SU, we investigate the power allocation strategies to maximize the ergodic sum rate of a fading C-MAC without successive interference cancelation (SIC). In particular, this paper considers two types of constraints: 1) average TPC and average IPC and 2) peak TPC and peak IPC. For the first case, it is proved that the optimal power allocation is dynamic time-division multiple access (D-TDMA), which is exactly the same as the optimal power allocation to maximize the ergodic sum rate of the fading C-MAC with SIC under the same constraints. For the second case, it is proved that the optimal solution must be at the extreme points of the feasible region. It is shown that D-TDMA is optimal with high probability when the number of SUs is large. Moreover, we show that when the SUs can be sorted in a certain order, an algorithm with linear complexity can be used to find the optimal power allocation.

Capacity Theorems for the Gaussian Zigzag Channel

We consider the two-user Gaussian zigzag channel, where there are two senders and two receivers. One of the senders transmits information to its intended receiver (without interfering with the unintended receiver) while the other sender transmits information to both receivers. We first establish achievable rates for the discrete memoryless zigzag channel under strong interference which can be extended to the Gaussian zigzag channel with strong and very strong crossover link gain (a2 ges 1). We also determine an outer bound for the Gaussian zigzag channel under strong and very strong crossover link gain (a2 ges 1). Moreover, we establish the capacity of the Gaussian zigzag channel under (strictly) strong crossover link gain (1 les a2 les 1 + P1)

The capacity region of a class of two-user degraded compound broadcast channels

Weingarten et al. established the capacity region for the two-user degraded compound broadcast channel where the realizations of each user exhibit a certain degradedness order. The degradedness order is defined through an additional fictitious user whose channel is stochastically degraded with respect to each realization from one set (the stronger receiver) while each realization from the other set (the weaker receiver) is stochastically degraded with respect to him. Rather than specifying a fictitious user, we consider a two-user degraded compound broadcast channel which is pair-wise degraded, i.e., each realization from the weaker receiver is stochastically degraded with respect to each realization from the stronger receiver. In this paper, we first prove the capacity region for a discrete memoryless class of this broadcast channels where the weaker receiver has only two possible realizations while there is an arbitrary number of possible realizations for the stronger receiver. Next, we consider the equivalent class of degraded compound MIMO Gaussian broadcast channels. To show that Gaussian inputs attain the capacity region, we prove a new extremal entropy inequality using a technique recently introduced by Geng and Nair that was used to resolve the capacity region of the two-user MIMO Gaussian broadcast channel with common and private information.

On lossy source coding with side information under the erasure distortion measure

We consider the Wyner-Ziv and two way source coding problems with the erasure distortion measure. We characterize the rate-distortion regions for these settings, when the source and side information satisfy a positivity condition. Using our results, we show that, contrary to recent conjectures in the literature, feedback from the decoder to the encoder does not reduce the overall rate required in the Wyner-Ziv setting with erasure distortion, when the positivity condition is satisfied. Finally, we extend our techniques to characterize the rate-distortion regions of two multiterminal source coding settings, the Heegard-Berger setting, and the cascade source coding setting, when the positivity condition is satisfied.

Network Coding for Intra-Cell Communications in OFDMA Networks

In this letter, we consider a new communication scenario in cellular networks, which is called intra-cell communications (ICC), for which user pairs within the same cell exchange information through the base station (BS). A network coding based protocol is proposed to assist ICC in OFDMA networks to improve the spectral efficiency. In the uplink, each user accesses the BS following the OFDMA principle, while in the downlink, the BS transmits to each of the user pairs using the OFDMA principle. To do so, the BS performs bit level XOR (BLX) operation on the information bits received from each user pair. The output is then transmitted back to the same user pair through subcarrier sharing. We formulate the joint uplink/downlink subcarrier assignment and power allocation problem to maximize the weighted sum rate. While the formulated problem is non-convex, an asymptotically optimal, yet low complexity algorithm is proposed. Simulation results are presented to verify that the proposed protocol outperforms subcarrier pairing and assignment schemes which require no subcarrier sharing in both uplink and downlink.

Sum-rate maximization for spectrum-sharing cognitive multiple access channels without successive interference cancellation

In this paper, the sum-rate of a cognitive multiple access channel (C-MAC) is studied, where a secondary network (SN) with multiple secondary users (SUs) transmitting to a secondary base station (SBS) shares the spectrum band with a primary user (PU). An interference power constraint (IPC) is imposed on the SN to protect the PU. Under the IPC and the individual transmit power constraint (TPC) imposed on each SU, we investigate the power allocation strategies to maximize the sum-rate of the C-MAC without successive interference cancellation (SIC). We prove that the optimal solution must be at the extreme points of the feasible region. We show that Dynamic Time Division Multiple Access (D-TDMA) is optimal with high probability when the number of SUs is large. Furthermore, we show through simulations that the optimal power allocation to maximize the sum-rate of the C-MAC with SIC is optimal or near-optimal for our setting when D-TDMA is not optimal.

The Capacity Region of the Class of Three-Receiver Gaussian MIMO Multilevel Broadcast Channels With Two-Degraded Message Sets

Nair and El Gamal established the capacity region of the three-receiver multilevel broadcast channel (MBC) with two-degraded message sets. For the three-receiver MBC with two-degraded message sets, the output at receiver 2 is a degraded version of the output at receiver 1. However, no order of degradedness is imposed on the output at receiver 3. The transmitter sends a common message to all three receivers and a private message to receiver 1. By considering a specific discrete-memoryless example, Nair and El Gamal showed that a direct extension of the Körner-Marton region (for the general two-receiver broadcast channel with degraded message sets) is strictly suboptimal. They also considered a three-receiver Gaussian product MBC and showed that, restricted to Gaussian inputs, the direct extension of the Körner-Marton region is again strictly suboptimal. However, whether Gaussian inputs are optimal remained unresolved. In this paper, we show that Gaussian inputs, along with time-sharing between rate points obtained with Gaussian inputs, achieve the capacity region of the three-receiver Gaussian multiple-input multiple-output MBC (this includes the three-receiver Gaussian product MBC considered by Nair and El Gamal) with two-degraded message sets. Our proof relies on the channel enhancement technique introduced by Weingarten as well as the perturbation approach employed by Liu and Viswanath.

The capacity region of some classes of parallel degraded broadcast channels with three receivers and three-degraded message sets

We consider a broadcast channel with three receivers and three-degraded message sets, where the transmitter has a common message intended for all three receivers, a message intended for receivers 2 and 3, and a private message intended for receiver 3. The messages are transmitted over a family of parallel degraded broadcast channels. In the most general case, there are six parallel degraded broadcast channels with different orders of degradedness. We determine the capacity region for two classes of broadcast channels, one with five parallel degraded broadcast channels and the other with three parallel degraded broadcast channels. The main difficulty is in identifying the relevant auxiliary random variables in the proofs of the converse. To accomplish this, we make use of an information theoretic inequality that can be proven using the Csiszár-sum identity.