Golnaz Farhadi

A Learning-Based Network Selection Method in Heterogeneous Wireless Systems

With the coexistence of various wireless technologies, next generation wireless communications will likely consist of an integrated system of networks, where the Access Points (APs) and Base Stations (BSs) work together to maximize the mobile-user Quality of Service (QoS). In such heterogeneous environment where handheld devices with different access technologies are not uncommon, it should be possible to select networks and seamlessly switch from one AP/BS to another in order to elevate user performance. In this paper, this type of network selection and handover mechanism with the goal of maximizing QoS is formulated as a Markov Decision Process (MDP). An algorithm based on Reinforcement Learning (RL) is then obtained that selects the best network based not only on the current network load but also the potential future network states. This algorithm aims at balancing the number of handovers and the achievable QoS. The results illustrate that while the QoS performance of the proposed algorithm is comparable to the performance of the optimum opportunistic selection algorithm, fewer number of network handovers (on average) are required. In addition, compared to the existing predefined network selection strategies with no handover, the MDP-based algorithm offers significantly better QoS.

Group-Based Signaling and Access Control for Cellular Machine-to-Machine Communication

Global connectivity and broad coverage of cellular networks constitute a platform for ubiquitous machine-to-machine (M2M) service provisioning on a large scale. However, the access attempt to the radio access network from a massive number of devices demands new cellular access mechanism and signaling design to avoid congestion of control channels. We propose group-based access and signaling policies through a two-level device partitioning, namely, paging groups and access groups. Group paging enables the base station to coordinate access to the network and hence control the random access channel (RACH) overload. However, unlike the conventional scheme where each individual device attempts access to the network, our proposed group-based access mechanism delegates the random access procedure of the devices in an access group to a designated device, referred to as group delegate. In addition, we develop signaling schemes that enable resource allocation at the level of access groups for direct data report as well as a group-based connection via group delegate. Numerical results show that the group access procedure achieves significantly smaller access delay than the conventional access scheme. It is also shown that the group-based signaling policies require signaling overhead proportional to the number of group delegates instead of the number of devices. The proposed group-based schemes enable scaling the number of devices paged simultaneously while controlling the access signaling congestion, access delay, and overhead. Furthermore, the group-based connection enables more resource re- use and hence more efficient resource usage for data transmission.

A Decentralized Power Allocation Scheme for Amplify-And-Forward Multi-Hop Relaying Systems

A decentralized power allocation scheme for amplify-and-forward (AF) multi-hop relaying systems is proposed. The scheme maximizes a harmonic mean-based approximate expression for the instantaneous received signal-to-noise ratio (SNR). The outage probability in Rayleigh fading of the proposed power-optimized system is evaluated. Numerical results show the superior performance of AF multi-hop relaying systems employing the proposed power allocation scheme over those with uniform power allocation. The asymptotic outage behavior for sufficiently large values of SNR is also studied and it is demonstrated that AF multi-hop relaying systems employing the proposed power allocation scheme achieve diversity order 2.

User Behavior Shift Detection in Ambient Assisted Living Environments

Identifying users’ frequent behaviors is considered a key step to achieving real, intelligent environments that support people in their daily lives. These patterns can be used in many different applications. An algorithm that compares current behaviors of users with previously discovered frequent behaviors has been developed. In addition, it identifies the differences between both behaviors. Identified shifts can be used not only to adapt frequent behaviors, but also shifts may indicate initial signs of some diseases linked to behavioral modifications, such as depression or Alzheimer’s. The algorithm was validated using datasets collected from smart apartments where five different ADLs (Activities of Daily Living) were recognized. It was able to identify all shifts from frequent behaviors, as well as identifying necessary modifications in all cases.

Tethering over TV White-Space: Dynamic Hotspot Selection and Resource Allocation

In dense wireless areas where the cellular spectrum resources are insufficient, the conventional approach is to install more base-stations (BS) or offload some of the traffic onto unlicensed WiFi bands. Both approaches require adding new infrastructure that might be necessary for only a short time. Taking advantage of dual-hop communication, frequency reuse policies, and opportunistic use of white-spaces, this paper proposes an algorithm for operator-controlled tethering over TV white-space (TVWS). As such, in a dense wireless area, some nodes can act as hotspots and tether data to and from their corresponding slaves over TVWS. The proposed algorithm iteratively clusters the nodes into hotspots and slaves, and allocates resources with the objective of maximizing spectrum utilization. Evaluations show that the proposed algorithm can effectively exploit TVWS spectra, and given a fixed amount of network resources, it can significantly increase the number of supported users.

CaSRA: An algorithm for cognitive tethering in dense wireless areas

This paper investigates the performance gain obtained by creating a hotspot-slave configuration of nodes in densely populated areas. In doing so, a semi-distributed algorithm, referred to as CaSRA (Clustering and Spectrum assignment and Resource Allocation), that allows the hotspots to tether over locally available white-spaces is proposed. CaSRA, performs in three-steps: 1) clusters the nodes based on the K-means clustering algorithm, 2) assigns white-space spectrum to each cluster based on a distributed graph-coloring approach to maximize spectrum reuse, and 3) allocates physical layer resources to individual users based on local channel information. Unlike small cells (femtocells, relays, and WiFi networks), this approach does not require any additions to the existing infrastructure, but allows the nodes, themselves, to act as hotspots. Simulation results show that given fixed amount of network resources, the proposed algorithm can significantly improve the overall performance of network users.

Dynamic handoff decision in heterogeneous wireless systems: Q-learning approach

The satisfaction of a mobile user in a heterogeneous wireless environment relies heavily on the appropriate choice of network. With the presence of various wireless technologies and advances in smart mobile devices, the mobile terminal in next generation wireless communications will likely make intelligent handoff decisions to optimize the user Quality of Experience (QoE). This paper investigates network selection and handoff decision with the goal of maximizing user QoE. An algorithm based on Q-learning is obtained that chooses the best network based not only on the current network state but also the potential future network and device states. As opposed to other dynamic programming-based algorithms, this method does not require the knowledge of the statistics of the wireless environment, but learns an optimum policy by utilizing the mobile devices past experience. It is shown that the QoE results of the proposed Dynamic Handoff Decision (DHD) algorithm come very close to the performance of an optimum oracle algorithm, while on average fewer number of network handoffs are required.

Spectral Efficient Multihop Relaying Based on Alternate Transmission

A spectral efficient multihop relaying scheme is proposed based on scheduling the source transmission in alternate time slots. Each relay processes its received signal and forwards it in the subsequent time slot. An interference cancellation mechanism is developed to eliminate the interference term at each relay caused by the source alternate-based transmission. The frame error probability and the achievable rate of the proposed scheme are evaluated. It is shown that the individual bit-by-bit detection at the destination performs almost as well as the optimal joint detection, and achieves a data rate that approaches the Shannon channel capacity limit. In addition, the proposed scheme significantly outperforms the conventional orthogonal multihop transmission in terms of the achievable data rate, especially for larger numbers of hops, with slightly inferior frame error rate performance.

Coordinated Tethering Over White Spaces

Coordinated tethering over a white-space spectrum is investigated herein to increase mobile broadband spectrum efficiency in densely populated areas. This paper proposes an algorithm for operator-controlled tethering over white spaces. The proposed approach does not add to the existing infrastructure but instead allows the individual nodes to act as “hotspots” and to tether data to and from other nodes. The proposed algorithm iteratively clusters the nodes into hotspots and slaves and allocates resources to maximize spectrum utility. The proposed methods dynamic characteristics allow cellular systems to hierarchically evolve in dense areas as necessary. A signaling framework for node-to-node and base-station-to-node communication that enables such operator-controlled tethering is also presented. Simulation results show that given a fixed amount of network resources, the proposed algorithm can significantly increase the number of supported users.

Optimal in-network cache allocation and content placement

We examine the in-network optimal content placement and storage allocation problem by formulating a linear program. Our objective is to minimize the overall cost of content delivery subject to total storage budget and link capacity constraints. The solution determines whether/where to keep a copy of a content based on contents ranking distributions. In addition, we present a realization of the optimal content placement solution based on content-centric networking protocol. We show that such optimized content delivery significantly reduces the cost of content distribution and improves quality of service.