On
Design of Next-Generation Wireless Metropolitan Area Networks With
ABSTRACT
Broadband wireless
access (BWA) defined in IEEE 802.16 is one of the most exciting advances in the
4G wireless communication revolutions for enabling the application scenario of
wireless metropolitan area networking, which has a great potential to
complement the current last-mile technologies, such as cable modem and digital
subscriber lines (DSL), for supporting user mobility and ubiquitous computing.
Instead of employing IEEE 802.16e mesh mode, Mobile Multi-hop Relay (MMR)
defined in IEEE 802.16j has been recognized as an economic yet effective
approach for coverage extension and throughput enhancement in IEEE 802.16d
Point-to-Multi-Point (PMP) networks, and has caught extensive attention from
both industry and academia for the design and implementation of the
next-generation WMANs. This project brings together a group of scientists,
research engineers, and graduate students, working closely with our industry
partners – Bell Mobility and SparkMatrix Technologies
Inc., to come up with an integrated and practically deployable
solution plane in such a promising network scenario. In specific, we are
committed to tackle some key issues in pushing the MMR networks into success,
including relay station allocation, cooperative communications, and routing/resource
allocation with mobility prediction under an integrated virtual-cell design. We
expect that the proposed schemes will not only facilitate our industry partners
in developing a practically implementable system and the corresponding
value-added services, but also contribute to the Internet and wireless
communications community in the state-of-the-art research toward new design
policies and standards for wireless metropolitan area networks (WMANs).
INTRODUCTION
The IEEE 802.16
standard (also known as WiMax) has attracted
extensive attentions from both industry and academia due to its economic and
interoperable characteristics, which provide an all-IP environment for
heterogeneous and multi-customer service support. The associated
implementations on wireless metropolitan area networks (WMANs) have been taken
as an effective complement to the legacy cable modem and Digital Subscriber
Line (DSL) technologies for broadband access. Compared
with the IEEE 802.11 technology (also known as Wifi),
IEEE 802.16 has a much larger coverage area and better potential capability for
Quality-of-Service (QoS) guarantee due to the
reservation-based resource allocation with OFDMA/TDMA. It is envisioned that
the deployment of IEEE 802.16 technology will serve as a major effort in the
evolution of wireless communication systems toward the fourth-generation (4G)
scenario.
Relaying is the most effective technique in extending
coverage and increasing capacity by reducing interference and the amount of
transmission power needed, especially when the communications are over long
distances. By adaptively and properly allocating relays into a metropolitan
area cellular network, the black holes can be effectively eliminated by way of
those relatively cheaper relaying devices. Furthermore, relays are more easily installed than base stations (BSs) due
to their smaller size, lower power consumption, and the elimination of backhaul
communication cable. The 802.16j Mobile Multi-hop Relay (MMR) working group has been working
extensively on exploring the possibility of incorporating relay stations (RSs)
with the conventional 802.16 network configuration. MMR aims to push the
success of the WMANs by extending coverage and enhancing throughput of the
legacy 802.16d Point-to-Multi-Point (PMP) networks.
Three types of relays are considered in the expansion
of 802.16 PMP mode: Fixed,
This project brings together a group of scientists, research engineers, and graduate students, working closely with our industry partners – Bell Mobility and SparkMatrix Technologies Inc., to develop an integrated and practically deployable solution plane in MMR networks. Due to the intrinsic multi-hop characteristic and mobility of both RSs and MSs, a number of unique issues and distinguishable requirements other than those by the legacy wireless communication systems (such as 3G mobile networks) have been raised in the design of the MMR networks, such as the adoption of advanced multi-user coding schemes, the support of versatile and heterogeneous types of service, the capability in service differentiation, and the requirement for high availability and service continuity for both non real-time and real-time services. Thus, this project aims to create a framework of control and management leading to the next-generation WMANs with 802.16j based MMRs. In particular, we are committed to investigating the RS placement problem and incorporation of advanced multi-user coding schemes for achieving high efficiency and physical availability. In terms of network management, the virtual cell concept will be studied and implemented in the MMR networks, where issues on intra-cell handover and high availability design will be explored. The outcomes of this research project will not only facilitate our industry partners in developing a practically implementable system and the corresponding value-added services, but also contribute to the Internet and wireless communications communities in the state-of-the-art research toward new design policies and standards for WMANs.
SECTION 1: RESEARCH GOALS,
METHODOLOGY, AND PLANNING
1.1 OBJECTIVES
Our long-term goal is to achieve a general framework that can enable an efficient, adaptive, and highly available WMAN environment with multi-customer and multi-service support. The proposed framework and solution plan are expected to solidly contribute to the research community and telecommunication industry in building up the next-generation WMANs. Our short-term objectives in the coming two years are as follows.
· To gain deeper understanding on the issues of relay placement, cooperative relaying, and virtual cell operations in 802.16j MMR networks in order to speed up the research and deployment progress of our research partners. In specific, the project will focus on the following three research topics:
- Optimal RS placement algorithms to ensure maximum MMR efficiency and the development of placement updating methods to allow timely adaptation to changes in network traffic demand.
- Optimal integration of cooperative relaying with network coding schemes, by exploiting both beamforming and binning techniques, with robustness to network topological variations.
- Development of a framework of intra-cell handover in order to construct a virtual cell environment for the MSs, and investigation into the mobility prediction based routing and resource allocation strategies by taking advantage of IEEE 802.21 standard signaling architect information exchange mechanisms.
· To generate new information and to disseminate knowledge through interactions and collaborations with the supporting company and publications in archival journals, premier conference proceedings, technical seminars, and invention disclosures and patents.
· To train graduate students and postdoctoral fellows in the aspects of Internet technology and wireless communication network design. The participants of the project will be benefited in terms of the knowledge and trainings on network relaying, coding, and efficient intra-cell control and management.
1.2 DETAILED RESEARCH PROPOSAL
1.2.1 Relay Station Placement
Mulithop relaying with heterogeneous
fixed, mobile, and nomadic stations is a defining characteristic of IEEE 802.16
MMR networks. For fully realizing the benefits of adopting RSs in a cell, a
central question is the optimal placement of RSs for efficient system
operation.
Early research into
relay-enabled wireless networks was mostly conducted within the context of
mobile ad hoc networks [1]. The more recent concept of multihop
cellular networking was a consequence of merging ad hoc networks and cellular
networks, using peer mobile hosts to relay data [2][3]. Furthermore, the
benefits of using immobile relay nodes have been studied to alleviate traffic
congestion in cellular networks [4] and in the mesh networking environment [5].
However, few existing studies in the literature have dealt with the optimal
placement of relay stations. In [6], one of the applicants and his research
team reported a Lagrangian iterative optimization and
solution framework to place relay nodes to maximize the network capacity of a
wireless LAN. Although this work provided insights into the optimal relay
placement problem, it was restricted to the LAN environment, allowing only
two-hop relaying and simple data forwarding. In another recent work [7] of the
applicants, some preliminary results on optimal RS placement were obtained for
different coding schemes with relatively simple relay structures. While
multi-hop relaying, with even more complicated multi-source multi-destination
structure, has been shown to significantly increase the total network
throughput [8][9],
the nature of multi-hop wireless connectivity poses unique challenges in
the optimal placement of RSs in an MMR network. Multi-hop routing and multi-hop
radio interference give new dimensions to the problem space, and the
integration of fixed and non-fixed RSs further complicates the design of relay
infrastructure. In this regard, we will focus on the following two vital components
of the RSs placement problem for MMR systems.
A Optimal
Placement of Multi-hop RSs
The optimal placement of multi-hop
RSs for data forwarding can be formulated as a mathematical program. The inputs
into this program may include the location candidate sites for the RSs, the
network user population and movement, and the traffic demand patterns. The
solution may also depends on the link capacities, the interference model, the
unit prices of networking equipment and communication links, and the system performance
requirements in terms of QoS and cost. The outputs of
interest may include the placement of RSs with additional designs such as
channel assignment and flow control. In location theory, this problem is
similar to the “hub location” and “maximum coverage” problems, with the
additional considerations of interference and channel assignment.
Unfortunately, this is a large optimization problem with a mixture of
continuous and discrete variables. In general, such a problem is NP-hard, and
there is no known efficient method to solve it. Explicit enumeration through
exhaustive search would lead to an exponentially increasing number of linear
programs, while the direct application of the conventional branch-and-bound
method for integer programs is equally intractable.
In this context, we
will investigate into efficient iterative solutions for the RS placement
problem. An iterative solution breaks down the original problem into a sequence
of smaller optimization sub-problems. Each sub-problem admits easier solutions.
When properly tuned, the sequence of solutions converges to an optimal or
near-optimal global solution. One common iterative approach is the Lagrangian constraint relaxation technique with gradient
descent search. However, this approach is not directly applicable to the RSs
placement problem due to the non-convex nature of the problem space. Hence, new
iterative optimization solutions are necessary for MMR networks.
The first step of our
proposed approach is to properly formulate the original mathematical program to
facilitate tractable solutions in the later steps. Since the same physical
properties and requirements of a large network can be expressed in different
ways mathematically, this step requires trials and errors, as well as insights
into the overall structure of relay placement in MMR systems. Transformations
including discretization on the one hand and smoothing on the other can be
applied to simply the original problem. Then, the mathematical program will be
decomposed into an iterative form, for example using a method similar to the
Bender decomposition, by gradually separating and re-applying the program
constraints. The uniqueness of such an approach is in its dependence on the
underlying mathematical structure given by the physical properties of the MMR
system. Finally, the iterations will be conducted as a sequence of simpler
optimization sub-problems, to be solved by standard techniques. Hence, a main
focus of our research will be on the identification of MMR system properties
that can be expressed in mathematical structures admitting both computationally
tractable decomposition and timely convergence.
B Relay
Station Placement Updating
Since the topology and data
traffic patterns of an MMR network change over time, RS placement must adapt to
such changes in order to maintain optimal system efficiency. The fixed RSs are
designed based on long-term characteristics of the network; their physical
placement is not expected to change, although the associative channel
assignments may change more often depending on large-scale traffic
fluctuations. In contrast, the nomadic RSs are designed to be movable according
to finer time-varying system demands. Therefore, it is vital to design fast RS
placement computation methods that allow timely adaptation. While an NP-hard
optimization problem of practical scale in MMR networks (tens to hundreds of
nodes) often requires days to weeks of computer time, we plan to investigate
into two approaches to reduce the computation complexity to allow hourly to daily
RS placement updates.
The first approach
involves novel application of heuristics. In each step of the aforementioned
iterative method, heuristics can be applied to reduce the vast computation
complexity. Heuristics by definition do not lead to optimality guarantees for
the problems they solve. However, there exist methods that guarantee the global
optimality of the final result even when the intermediate iterative steps yield
sub-optimal results. We plan to investigate into these methods, and we will further
study the tradeoff between optimality and computation complexity. The second
approach is the continuous maintenance of RS placement. We will develop
placement solutions that require only incremental computation given the
previous placement solution. When the network traffic demands change, instead
of re-computing the optimal RS placement for an entire network, the framework
of continuous maintenance will take the previous solution and the ensued
traffic changes as input to quickly compute a new solution. This will be
enabled by an extension to the aforementioned iterative method that adaptively
changes the direction of search in the solution space based on changing system
parameters. The proposed fast updating approaches will be applied to both fixed
RSs and nomadic RSs, which have different levels of stability and hence
different computation updating frequencies. Therefore, the flexibility to
choose various levels of optimality, and hence the corresponding speeds of
computation, will be an essential feature of the proposed research outcome.
1.2.2 Cooperative Relaying with Network Coding
The research
on relay coding schemes at the physical layer has been an old and well-studied
topic [10][11]. There are two basic strategies:
decode-and-forward, and compress-and-forward. Some recent research has been
able to extend these basic schemes to large wireless networks with multiple
relays [8][9][12]. Since relay schemes exploit all the
useful signals instead of treating them as interference, it becomes a natural
candidate for the more advanced operation of wireless networks.
Although a relatively new research topic, network coding has attracted a lot of research interests in recent years. Besides the wide variety of potential applications, the simplicity of the essential idea of network coding also contributes to its success. Currently, one major research focus on network coding is its application to wireless networks. Some preliminary results [13] show that not only the idea of network coding is applicable to wireless networks, but also the broadcast nature of wireless communication is especially suitable for the implementation of network coding. More interestingly, some recent discovery [14] indicates that relaying and network coding can be naturally combined into the same framework for wireless communication, and they together form a simple and easily applicable structure, in a way both the relay gain and network coding gain can be achieved simultaneously.
Nevertheless, the incorporation between relaying and network coding presents both unforeseen opportunities and challenges. In order to fully realize the promised gains, some fundamental challenges need to be resolved. In this project, the following research issues will be addressed.
A Beamforming
vs. Binning
Beamforming is an effective technique to increase the transmission rate by boosting
the received signal power. As a basic characteristic of many cooperative
communication schemes, beamforming can be easily
exploited with the decode-and-forward relaying scheme, where different
transmitters can fully cooperate based on common information.
Binning
is one of the most classical and fundamental coding techniques in network
information theory. By binning different messages into the same representative
codeword according to the network structure, the same end-to-end performance
can be achieved with lower communication rates. This can be extremely useful
when multiple communication tasks are competing for the limited resources in a
single communication network. Actually, the binning idea of letting one
transmission help many is exactly the essential idea of network coding. It was
shown in [14] that in wireless networks, the natural way to realize network
coding is by the binning technique.
Since beamforming and binning are beneficial in different
perspectives, it is of great interest to exploit both of them in our coding
schemes. However, their combination cannon be easily realized, due to the
intrinsic conflict between their basic operational principles. In order to
carry out beamforming, the transmitters must have
common information to send, while binning is useful only when independent
messages are mixed. Hence, there is a basic tradeoff on choosing common or
different information to transmit. Considering their relative gains and cost,
it is of fundamental importance to characterize the optimal operational point.
Our approach to this
problem starts with the basic two-way relay channel, which models both the
downlink and uplink of a single relay station scenario. We have obtained some
preliminary results on this simple case in [14], which, together with the
insight gained from our previous work [8][9] on the
multiple-relay networks, will constitute the basis for our further exploration
of the general scenarios where multiple relay stations are present. The final
results are expected to provide fundamental insight and practical guidance to
the basic tradeoffs between coding gains, computational complexities,
implementation cost, as well as the related RSs placement problem mentioned
before.
B Topological Robustness
Apparently, the optimal
communication strategy to choose depends on the network topology under design.
It is well known that even for the simplest one relay channel [11],
dramatically different coding schemes, namely, decode-and-forward and
compress-and-forward, are called for different scenarios. Even more critically,
network coding schemes are designed almost completely based on network
topology. Since frequent switching between dramatically different coding
schemes is not a cost-effective choice, the successful application of a coding
scheme to mobile wireless networks requires not only its adaptability to major
topological changes, but also its robustness to small variations.
In fact, a study of
the three-way relay network [14] indicates that the optimal combination of
relaying and network coding critically depends on the network topology.
Therefore, in order to achieve robust performance, it is highly desirable to
design coding strategies together with network topology, which in turn is
closely related to the RSs placement problem discussed before. In this regard,
we will characterize the robustness regions of different coding schemes, by
studying the performance under various topological perturbations.
1.2.3 Virtual Cell Design and Implementation
The concept of virtual BS (or called virtual cell) has emerged as one of the ultimate design goals for the scenario of MMR networking, which is expected to be in a practical implementation of the next-generation Wireless MANs. The concept of virtual cell did not attract much attention in the traditional cellular networks of single hop connections in nature. In 802.16 MMR networks, since each RS has its own coverage range which may overlap with that of the others, some MSs could be connected to multiple RSs in meeting the high availability design requirement. In addition, due to the intrinsic nature of multi-hop connections in the MMR networks, simply performing handover based on the decision by the MS and/or assisted by the local RSs associated with the MS may lead to inefficient use of network resources possibly caused by triangle routing problems and unbalanced resource utilization.
Since the RSs are taken as cheap devices, we envision that the number of BSs in a cell could be large, while the coverage of each RS could be small and possibly overlapped in order to improve service availability and eliminate black holes. Also, the BS will be the major network entity with the intelligence of coordinating the RSs, such that the maximum extent of frequency reuse and synchronization can be achieved through a well designed scheduling scheme and real-time signaling. Furthermore, the standard of IEEE 802.16j has each RS supporting fast route change such that a cell-wide path rerouting process for a single MS can be easily implemented. We expect for a high frequency of handover events between adjacent RSs not only due to the mobility of MSs and RSs, but also any possible reconfiguration process which change routes for some existing connections. Therefore, instead of migrating the previously reported schemes for the legacy cellular and mobile ad hoc networks, it is highly desired to freshly develop a suite of interoperable strategies for intra-cell handover for achieving load balancing and seamless handover. Also, due to the unreliability wireless channels and cheap RS devices, we may need to come up with a strategy of redundancy placement for improving service availability inside a cell, which is essential in supporting the highly bandwidth demanding and interruption-sensitive applications.
A Intra-Cell Handover
To the best of our survey, there is no study that has investigated into BS assisted intra-cell handover by considering the fact that the RSs are highly synchronized, where the handover decision is most likely made by the MSs. In this project, we will explore the concept of virtual cell operation in MMR networks so as to take the best advantage of demonstrated unique features distinguished from the legacy wireless networks, such as mobility of RSs, support of fast route change, and the carrier-owned infrastructure with multi-hop connections. In particular, we are committed to developing an efficient intra-cell handover strategy in order to handle the emerging real-time and highly interactive services with stringent delay constraints. The ultimate design objective is to achieve a seamless integration of RSs in a cell in terms of minimized handover delay, efficient resource allocation and management efforts, such that each MS can be totally unaware of the existence of the RSs and the handover activities among them. For achieving this, the BS has to trace each MS by way of the signaling strength reported from the RSs and set up new paths for each MS when pre-defined handover conditions are met. In addition, some mobility prediction mechanisms for the MSs must be equipped at the BS when handover among the RSs is performed.
In the project, instead of making all the decisions at the MS (which is the basic assumption taken by most of the previously reported studies), we will investigate into the scenario of network initiated handover, where the BS notifies the foreign RSs to start pre-handover (i.e., the setup of a new link from the foreign RS to the MS). In this case, the handover of the MS within a cell will be transparent to the MS. Specifically, each MS in the cell will first be connected to the BS, and the BS will then determine if there is a RS with a better communication channel to provision the service. If yes, the BS will hand over to the RS, or cooperatively provision the bandwidth through a decode-forwarding scheme. When the MS is moving out of the range of the RS (while still in the cell), the BS will re-select an adjacent RS with the best communication channel to take over the connection. To help the selection of adjacent RSs for a MS, the following two tasks will be investigated:
Motivated by the fact that the user mobility is mainly dominated by the geographic arrangement of the hotspot, such as streets, hotspots, and buildings/constructions, we propose a Markov chain based parameter training mechanism at the BS to gain the knowledge of user mobility pattern in the coverage of each RS [15], in order to improve the accuracy in user mobility prediction. Since the obtained parameters through the training process are specific to the geographic arrangement of the RS coverage, we expect a much better accuracy in achieving the desired performance in the vertical handoff process can be yielded. In addition, since the pre-handover decision is made at the BS, the design complexity and power consumption of the MSs can be significantly reduced. We will justify that the proposed mechanism can be easily performed under the service supports of the 802.21 framework [16]. As one of the most distinguished features from the previously reported counterparts, the performance of the proposed scheme will be independent of which user mobility model is assumed since the BS derives the user mobility pattern through the historical user mobility record. We will conduct extensive analysis for modeling the system performance in terms of error propagation from the signal detection inaccuracy to the induced error in the measured handoff delay.
In addition to the user mobility portrait, the handover decision and neighbor RS selection will try to optimize load balancing and capacity efficiency according to the current network states. We will firstly define the capacity of each RS and link availability for each pair of RSs according to its historical channel condition to the neighbor RSs and BS as well as the interference among RSs due to the multi-hop relaying. By jointly considering the macro-mobility prediction, capacity of each RS, network topology and routing strategy, and link availability for each pair of RSs, a suite of cost functions and preference metrics for route selection will be obtained based on the RS capacity. Based on the costs function and preference metrics, an intra-cell handover decision algorithm for dynamic route selection will be developed and functioning at the BS in order to supervise each active MS within the cell.
B Survivable Routing and Redundancy Planning
It has been well recognized that RSs are relatively cheap devices with abundant capacity yet less reliability, while the BS serves as the capacity bottleneck in the cell. Thus, the capacity of RSs could be much under-utilized. Our design will take advantages of the under-utilized RSs and rearrange them into effective spare capacity in the course of bandwidth provisioning. We aim to achieve a high-availability design through intra-cell redundancy planning efforts in terms of survivable and multi-path routing.
First of all, our design has every MS to be always associated with two different RSs, and the network topology to be at least 2-connected. This is more than possible in the scenario of multi-hop relaying, where the coverage ranges of RSs should be highly overlapped. Since the RSs form a meta-mesh network such that more than one path exists between a RS and the BS, our design has two node-disjoint paths between a MS and the BS to be routed in order to improve intra-cell service availability for each MS. The primary path provisions the bandwidth for the MS, which is composed of one or a number of intermediate RSs; while the RSs along the secondary path are pre-selected with the capacity being reserved. This is also referred to as 1+1 protection which aims to achieve the highest intra-cell service availability. Instead of 1+1 protection, shared protection can be employed for exploring better advantages due to the multicasting in nature communications and achieving better capacity efficiency at the expense of less service availability. In this case, multiple primary paths can share the common capacity as the corresponding redundancy. In other words, a portion of capacity of a RS is reserved as a pool for all the associated primary paths, and as long as any primary is interrupted which activates the restoration mechanism, the BS will activate the secondary path through the RS for restoring the interrupted service.
It is clear that the
network environment could be changed not only due to traffic distribution
variation, but also due to the change of network topology, long-term channel
conditions among RSs, and any possible QoS
requirement. Thus, in order to cope with the network environment change, we
will develop a route reconfiguration strategy by investigating tree-based fast
rerouting for the data paths of each RS. The formation of colored trees (Red
and Blue trees) [17] will be considered to route data and can provide flexible
recovery in the topology. In order for such trees to be built, the original
relay meta-mesh network must be 2-connected. The routing tree structures will
be constructed within the relay mesh topology. This will be achieved as
described as follows: We will develop a centralized routing algorithm similar
to that of [17] by taking into consideration wireless characteristics,
specifically the link metrics defined in the project, for achieving load
balancing, interference avoidance, and frequency reuse, according to the given traffic
demand distribution. Mathematical models to incorporate the proposed link
metrics into the routing algorithm will be needed. Thus, we will develop an availability
constrained survivable routing approach using the tree techniques. We will also
investigate fast reconfiguration of the trees in order to cope with the traffic
demand variation.
1.3 METHODOLOGY
The proposed
schemes, algorithms, and approaches will be formulated and modeled through
mathematical formulation, and illustrated and validated through simulation.
Taking advantages of the optimization tools (CPLEX and Matlab)
and high-end workstations derived using the funds from the Canadian Foundation
for Innovation (CFI), we expect to develop large-scale
network performance simulators for examining the developed approaches on
achievable throughput. Due to the strong cooperative relationship with our
industry partners, we will launch the developed schemes and models in their testbeds and practically operate the proposed schemes.
Therefore, the proposed research program will be pursued using all available
analytical, simulation, and experimentation approaches.
REFERENCES
[1] C. E. Perkins, Ed., Ad Hoc Networking,
Addison-Wesley Longman, 2001.
[2] Y. Lin and Y.
Hsu, “Multihop cellular: A new architecture for wireless
communications,” in Proc. of IEEE INFOCOM, Tel-Aviv, Israel, March 2000,
vol. 3, pp. 1273 – 1282.
[3] F. Fitzek, D. Angelini,
G. Mazzini, and M. Zorzi, “Design and performance of
an enhanced IEEE802.11 MAC protocol for ad hoc networks and coverage extension
for wireless networks,” IEEE Wireless Communications, vol. 10,
no. 6, pp. 30 – 39, Dec. 2003.
[4] H. Wu, C. Qiao, S. De, and O. Tonguz,
“Integrated cellular and ad hoc relaying systems: iCAR,”
IEEE Journal on Selected Areas in Communications, vol. 19, no.
10, pp. 2105–2215, Oct. 2001.
[5] Y. Bejerano, “Efficient integration of multihop wireless and wired networks with QoS constraints,” IEEE Transactions on Networking,
vol. 12, no. 6, pp. 1064– 1078, Dec. 2004.
[6] So and B. Liang, “Enhancing WLAN capacity by strategic placement of tetherless relay points,” IEEE Transactions on Mobile
Computing, vol. 6, no. 5, pp. 522-535, May 2007.
[7] B. Lin,
P.-H. Ho, L.-L. Xie, and X. Shen, "Optimal Relay Station Placement in IEEE
802.16j Networks", IWCMC 2007,
[8] L.-L. Xie and P. R. Kumar, ``An achievable rate for the multiple-level relay channel''. IEEE Transactions on Information Theory, vol.51, pp.1348-1358, April 2005.
[9] L.-L. Xie and P. R. Kumar, ``Multi-source, multi-destination, multi-relay wireless networks''. To appear in IEEE Transactions on Information Theory, Special Issue on Models, Theory and Codes for Relaying and Cooperation in Communication Networks, October 2007.
[10] E. C. Van der Meulen, “Three-terminal communication channels,” Adv. Appl. Probab., vol.3, pp.120-154, 1971.
[11] T. Cover and A. El Gamal, “Capacity theorems for the relay channel,” IEEE
Trans. Inform.
Theory, vol.25, pp.572-584,
1979.
[12] G. Kramer, M. Gastpar, and P. Gupta, “Cooperative strategies and capacity theorems for relay networks'', IEEE Transactions on Information Theory, vol.51, pp.3037-3063, Sep. 2005.
[13] Y. Wu, P. A. Chou, and S.-Y. Kung, “Information exchange in wireless networks with network coding and physical-layer broadcast,” in Proc. 2005 Conf. Inf. Sci. and Syst., Mar. 2005.
[14]
L.-L. Xie, “Network coding and random binning for multi-user
channels,” in Proc. of the 10th Canadian Workshop on Information Theory,
pp.85-88,
[15] S. –Y. Chang^ and Pin-Han Ho, “A Cooperative Two-Step Vertical Handoff Scheme with Mobility Prediction”, accepted (with minor revision) in IEEE Transactions on Vehicular Technology (in May 2007, 12 pages)
[16] Draft IEEE Standard for Local and Metropolitan Area Networks: Media Independent Handover Services: IEEE P802.21/D03.00, Std., Dec. 2006.
[17] M. Medard et al. “Redundant Trees for Preplanned Recovery in
Arbitrary Vertex-Redundant or Edge-Redundant Graphs,” in IEEE Journal of
Transactions on Networking, vol. 7, No. 5, October 1999, pgs. 641-652.
[18] P. Thulasiraman,