This paper presents a low-power spiking neural network (SNN) system enabling on-chip calibration with pulse-driven computation (PDC) for weight update based on the delta rule algorithm. During the calibration phase, the proposed PDC scheme computes delta weights without multipliers, thereby avoiding memory access overhead. By converting error values and input spikes into pulse-width and frequency signals, respectively, the proposed calibration architecture implements weight updates via simple pulse counting using counters and logic gates. Additionally, a weak softmax approximation and input scaling method are employed to reduce bit-width and maintain accuracy. The proposed SNN system, fabricated in a 28 nm CMOS technology, achieves an inference energy efficiency of 0.2 pJ/SOP and a calibration efficiency of 3.25 TOPS/W. Measurements on the MNIST dataset confirm that the proposed SNN system effectively compensates for process variations across multiple chips, achieving an accuracy improvement of 17–37% and a power reduction of 95–99% compared to conventional multiplier-based MAC designs.

Large language models (LLMs) have shown remarkable success, but aligning them with human preferences remains a core challenge.As individuals have their own, multi-dimensional preferences, recent studies have explored multidimensional personalization, which aims to enable models to generate responses personalized to explicit preferences.However, human preferences are often implicit and thus difficult to articulate, limiting the direct application of this approach.To bridge this gap, we propose Active Multi-dimensional Preference Learning (AMPLe), designed to capture implicit user preferences from interactively collected comparative feedback.Building on Bayesian inference, our work introduces a modified posterior update procedure to mitigate estimation bias and potential noise in comparisons.Also, inspired by generalized binary search, we employ an active query selection strategy to minimize the number of required comparisons by a user.Through theoretical analysis and experiments on language generation tasks, we demonstrate feedback efficiency and effectiveness of our framework in personalizing model responses.

In this paper, we propose a Spiking Neural Network (SNN) system using a proposed delta rule algorithm-based calibration method to improve accuracy during hardware implementation. The proposed weight calibration cell employs the pulse-driven computation (PDC) method rather than multiply-accumulation (MAC) to perform low-power operations. The PDC organizes the operational part as a counter that performs multiplication operations by counting the number of pulses over a specified time. Compared with existing MAC-based cells, the proposed weight calibration cell enables real-time weight updates without requiring additional memory access, as the counter is capable of performing both computational and memory functions concurrently. This approach results in a significant energy reduction of 95% to 99% and delivers a high energy efficiency of 155 pJ/Train. Additionally, the system achieves stable calibration accuracy, ranging from 94% to 95%, even across diverse accuracy environments.

Multi-Agent Path Finding (MAPF) is an important optimization problem underlying the deployment of robots in automated warehouses and factories. Despite the large body of work on this topic, most approaches make heavy simplifications, both on the environment and the agents, which make the resulting algorithms impractical for real-life scenarios. In this paper, we consider a realistic problem of online order delivery in a warehouse, where a fleet of robots bring the products belonging to each order from shelves to workstations. This creates a stream of inter-dependent pickup and delivery tasks and the associated MAPF problem consists of computing realistic collision-free robot trajectories fulfilling these tasks. To solve this MAPF problem, we propose an extension of the standard Prioritized Planning algorithm to deal with the inter-dependent tasks (Interleaved Prioritized Planning) and a novel Via-Point Star (VP*) algorithm to compute an optimal dynamics-compliant robot trajectory to visit a sequence of goal locations while avoiding moving obstacles. We prove the completeness of our approach and evaluate it in simulation as well as in a real warehouse.

BACKGROUND: E. coli chemotactic motion in the presence of a chemonutrient field can be studied using wet laboratory experiments or macroscale-level partial differential equations (PDEs) (among others). Bridging experimental measurements and chemotactic Partial Differential Equations requires knowledge of the evolution of all underlying fields, initial and boundary conditions, and often necessitates strong assumptions. In this work, we propose machine learning approaches, along with ideas from the Whitney and Takens embedding theorems, to circumvent these challenges. RESULTS: Machine learning approaches for identifying underlying PDEs were (a) validated through the use of simulation data from established continuum models and (b) used to infer chemotactic PDEs from experimental data. Such data-driven models were surrogates either for the entire chemotactic PDE right-hand-side (black box models), or, in a more targeted fashion, just for the chemotactic term (gray box models). Furthermore, it was demonstrated that a short history of bacterial density may compensate for the missing measurements of the field of chemonutrient concentration. In fact, given reasonable conditions, such a short history of bacterial density measurements could even be used to infer chemonutrient concentration. CONCLUSION: Data-driven PDEs are an important modeling tool when studying Chemotaxis at the macroscale, as they can learn bacterial motility from various data sources, fidelities (here, computational models, experiments) or coordinate systems. The resulting data-driven PDEs can then be simulated to reproduce/predict computational or experimental bacterial density profile data independent of the coordinate system, approximate meaningful parameters or functional terms, and even possibly estimate the underlying (unmeasured) chemonutrient field evolution.

We consider a multi-armed bandit problem in which a set of arms is registered by each agent, and the agent receives reward when its arm is selected. An agent might strategically submit more arms with replications, which can bring more reward by abusing the bandit algorithm’s exploration-exploitation balance. Our analysis reveals that a standard algorithm indeed fails at preventing replication and suffers from linear regret in time $T$. We aim to design a bandit algorithm which demotivates replications and also achieves a small cumulative regret. We devise Hierarchical UCB (H-UCB) of replication-proof, which has $O(\ln T)$-regret under any equilibrium. We further propose Robust Hierarchical UCB (RH-UCB) which has a sublinear regret even in a realistic scenario with irrational agents replicating careless. We verify our theoretical findings through numerical experiments.

IPTV service providers offering Video-on-Demand currently use servers at each metropolitan office to store all the videos in their library. With the rapid increase in library sizes, it will soon become infeasible to replicate the entire library at each office. We present an approach for intelligent content placement that scales to large library sizes (e.g., 100 Ks of videos). We formulate the problem as a mixed integer program (MIP) that takes into account constraints such as disk space, link bandwidth, and content popularity. To overcome the challenges of scale, we employ a Lagrangian relaxation-based decomposition technique combined with integer rounding. Our technique finds a near-optimal solution (e.g., within 1%-2%) with orders of magnitude speedup relative to solving even the linear programming (LP) relaxation via standard software. We also present simple strategies to address practical issues such as popularity estimation, content updates, short-term popularity fluctuation, and frequency of placement updates. Using traces from an operational system, we show that our approach significantly outperforms simpler placement strategies. For instance, our MIP-based solution can serve all requests using only half the link bandwidth used by least recently used (LRU) or least frequently used (LFU) cache replacement policies. We also investigate the tradeoff between disk space and network bandwidth.

When traffic arrives from the network for an idled mobile device, the network executes device activation procedures to wake the device up. Current device activation mechanisms are ill suited to support the expected growth of machine-to-machine (M2M) devices and traffic. We propose an adaptive device activation architecture for LTE/EPC cellular networks that adapts to network conditions and M2M application requirements to realize scalable device activation without increasing the resources used for this purpose. Our evaluation shows that our adaptive approach enables the network to handle M2M applications with a large number of devices without negatively impacting existing human-to-human (H2H) and human-to-machine (H2M) traffic.

Today's web page download process is ill suited to cellular networks resulting in high page load times and radio energy usage. While there have been notable prior attempts at tackling the challenge with assistance from proxies (cloud), achieving a responsive and energy efficient browsing experience remains an elusive goal. In this paper, we make a fresh attempt at addressing the challenge by proposing PARCEL. PARCEL splits functionality between the mobile device and the proxy based on their strengths, and in a manner distinct from both traditional browsers and existing cloud-heavy approaches. We conduct extensive evaluations over an operational LTE network using a prototype implementation of PARCEL. Our results show that PARCEL reduces page load times by 49.6%, and radio energy consumption by 65% compared to traditional mobile web browsers. Further, our results show PARCEL continues to perform well under client interactions, owing to its judicious functionality split.

Distributed clouds offer a choice of data center locations to application providers to host their applications. In this paper we consider distributed clouds that host virtual desktops(VDs) which are then accessed by their users through remote desktop protocols. VDs have different sensitivities to latency, primarily determined by the types of applications running (games or video players are more sensitive to latency) and the end users' locations. We design VMShadow, a system to automatically optimize the location and performance of latency-sensitive VDs in the cloud. VMShadow performs black-box fingerprinting of a VM's network traffic to infer its latency-sensitivity and employs a greedy heuristic based algorithm to move highly latency-sensitive VMs to cloud sites that are closer to their end users. VMShadow employs WAN-based live migration and a new network connection migration protocol to ensure that the VM migration and subsequent changes to the VM's network address are transparent to end-users. We implement a prototype of VMShadow in a nested hypervisor and demonstrate its effectiveness for optimizing the performance of VM-based desktops in the cloud. Our experiments on a private and the public EC2 cloud show that VMShadow is able to discriminate between latency-sensitive and insensitive desktop applications and judiciously move only those VMs that will benefit the most. For desktop VMs with video activity, VMShadow improves VNC's refresh rate by 90%. Further our connection migration proxy, which utilizes dynamic rewriting of packet headers, imposes a rewriting overhead of only 13μs per packet. Trans-continental VM migrations take about 4 minutes.

We present VMShadow, a system that automatically optimizes the location and performance of applications based on their dynamic workloads. We prototype VMShadow and demonstrate its efficacy using VM-based desktops in the cloud as an example application. Our experiments on a private cloud as well as the EC2 cloud, using a nested hypervisor, show that VMShadow is able to discriminate between location-sensitive and location-insensitive desktop VMs and judiciously moves only those that will benefit the most from the migration. For example, VMShadow performs transcontinental VM migrations in ~ 4 mins and can improve VNC's video refresh rate by up to 90%.

As cloud data services proliferate, filtering the communication between different virtual machines in a data center becomes a necessity. Such filtering can be accomplished by placing firewalls at strategic nodes within the data center network and rerouting the communication flows to pass through a firewall. This abstraction introduces several basic location problems which arise in these contexts. Suppose a VM s wishes to send data to a VM t along path P. If there is no available firewall on path P, we need to reroute the data first from s to a firewall f and then from f to the destination t. Clearly, having too few firewalls would cause a large number of communication flows to be routed to a particular firewall leading to increased congestion in the links leading to the firewall. As latency in data centers is dominated by link congestion rather than distance, we focus on finding good firewall placements subject to a bandwidth constraint on links.

We propose Joint-Family, a protocol that combines peer-to-peer (P2P) and adaptive bitrate (ABR) streaming for video-on-demand (VoD). While P2P for VoD and ABR have been proposed previously, they have not been studied together because they attempt to tackle problems with seemingly orthogonal goals. We motivate our approach through analysis that overcomes a misconception resulting from prior analytical work, and show that the popularity of a P2P swarm and seed staying time has a significant bearing on the achievable per-receiver download rate. Specifically, our analysis shows that popularity affects swarm efficiency when seeds stay “long enough”. We also show that ABR in a P2P setting helps viewers achieve higher playback rates and/or fewer interruptions. We develop the Joint-Family protocol based on the observations from our analysis. Peers in Joint-Family simultaneously participate in multiple swarms to exchange chunks of different bitrates. We adopt chunk, bitrate, and peer selection policies that minimize occurrence of interruptions while delivering high quality video and improving the efficiency of the system. Using traces from a large-scale commercial VoD service, we compare Joint-Family with existing approaches for P2P VoD and show that viewers in Joint-Family enjoy higher playback rates with minimal interruption, irrespective of video popularity.

Peer-to-Peer (P2P) systems have evolved from being used for file sharing to delivering streaming video on demand (VoD). The policies adopted in P2P VoD, however, have not taken user viewing behavior - that users abandon videos - into account. We show that abandonment can result in increased interruptions and wasted resources. As a result, we reconsider the set of policies to use in the presence of abandonment. Our goal is to balance the conflicting needs of delivering videos without interruptions while minimizing wastage. We find that an Earliest-First chunk selection policy in conjunction with the Earliest-Deadline peer selection policy allows us to achieve high download rates. We take advantage of abandonment by converting peers to “partial seeds”; this increases capacity. We minimize wastage by using a playback lookahead window. We use analysis and simulation experiments using real-world traces to show the effectiveness of our approach.

Broadcasting is a fundamental operation in wireless networks and plays an important role in the communication protocol design. In multihop wireless networks, however, interference at a node due to simultaneous transmissions from its neighbors makes it nontrivial to design a minimum-latency broadcast algorithm, which is known to be NP-complete. We present a simple 12-approximation algorithm for the one-to-all broadcast problem that improves all previously known guarantees for this problem. We then consider the all-to-all broadcast problem where each node sends its own message to all other nodes. For the all-to-all broadcast problem, we present two algorithms with approximation ratios of 20 and 34, improving the best result available in the literature. Finally, we report experimental evaluation of our algorithms. Our studies indicate that our algorithms perform much better in practice than the worst-case guarantees provided in the theoretical analysis and achieve up to 37 percent performance improvement over existing schemes.

Recently, practical subframe-level schemes, such as frame combining and partial packet recovery, have been proposed for combating wireless transmission errors. These approaches depend heavily on the bit error behavior of wireless data transmissions, which is overlooked in the literature. We study the characteristics of subframe bit errors and their location distribution by conducting extensive experiments on several IEEE 802.11 WLAN testbeds. Our measurement results identify three bit error patterns: slope-line, saw-line, and finger. Among these three patterns, we have verified that the slope-line and saw-line are present in different physical environments and across various hardware platforms. However, the finger pattern does not appear on some platforms. We discuss our current hypotheses for the reasons behind these bit error patterns and how identifying these patterns may help improve the robustness of WLAN transmissions. We believe that identifiable bit error patterns can potentially introduce new opportunities in channel coding, network coding, forward error correction (FEC), and frame combining.

IP Anycast has many attractive features for any service that involve the replication of multiple instances across the Internet. IP Anycast allows multiple instances of the same service to be “naturally” discovered, and requests for this service to be delivered to the closest instance. However, while briefly considered as an enabler for content delivery networks (CDNs) when they first emerged, IP Anycast was deemed infeasible in that environment. The main reasons for this decision were the lack of load awareness of IP Anycast and unwanted side effects of Internet routing changes on the IP Anycast mechanism. In this article we re-evaluate IP Anycast for CDNs by proposing a load-aware IP Anycast CDN architecture. Our architecture is prompted by recent developments in route control technology, as well as better understanding of the behavior of IP Anycast in operational settings. Our architecture makes use of route control mechanisms to take server and network load into account to realize load-aware Anycast. We show that the resulting redirection requirements can be formulated as a Generalized Assignment Problem and present practical algorithms that address these requirements while at the same time limiting connection disruptions that plague regular IP Anycast. We evaluate our algorithms through trace based simulation using traces obtained from a production CDN network.

IPTV service providers offering Video-on-Demand currently use servers at each metropolitan office to store all the videos in their library. With the rapid increase in library sizes, it will soon become infeasible to replicate the entire library at each office. We present an approach for intelligent content placement that scales to large library sizes (e.g., 100Ks of videos). We formulate the problem as a mixed integer program (MIP) that takes into account constraints such as disk space, link bandwidth, and content popularity. To overcome the challenges of scale, we employ a Lagrangian relaxation-based decomposition technique combined with integer rounding. Our technique finds a near-optimal solution (e.g., within 1-2%) with orders of magnitude speedup relative to solving even the LP relaxation via standard software. We also present simple strategies to address practical issues such as popularity estimation, content updates, short-term popularity fluctuation, and frequency of placement updates. Using traces from an operational system, we show that our approach significantly outperforms simpler placement strategies. For instance, our MIP-based solution can serve all requests using only half the link bandwidth used by LRU or LFU cache replacement policies. We also investigate the trade-off between disk space and network bandwidth.

Network providers face a huge challenge of running their network without service disruption in the presence of constant network change. Such change often involves router configuration update. The goal of gold standard auditing is to ensure all field configs are equivalent to a certified gold config. To handle constant change involving many features configured for numerous routers at the network edge, we need a scalable system that can perform gold standard auditing in a timely fashion. We present the design of GSAT (Gold Standard Audit Tool) that utilizes the syntactic structure of configs. When there is difference between gold and field configs, GSAT can provide users with enough structural hints to understand the context of the difference. Being technology-independent, GSAT is highly scalable and extensible. We have been using GSAT to audit customer-facing routers in a large-scale operational IP network. We present the implementation details and audit performance of our system.

A Shrew attack, which uses a low-rate burst carefully designed to exploit TCP's retransmission timeout mechanism, can throttle the bandwidth of a TCP flow in a stealthy manner. While such an attack can significantly degrade the performance of all TCP-based protocols and services including Internet routing (e.g., BGP), no existing scheme clearly solves the problem in real network scenarios. In this paper, we propose a simple protection mechanism, called SAP (Shrew Attack Protection), for defending against a Shrew attack. Rather than attempting to track and isolate Shrew attackers, SAP identifies TCP victims by monitoring their drop rates and preferentially admits those packets from the victims with high drop rates to the output queue. This is to ensure that well-behaved TCP sessions can retain their bandwidth shares. Our simulation results indicate that under a Shrew attack, SAP can prevent TCP sessions from closing, and effectively enable TCP flows to maintain high throughput. SAP is a destination-port-based mechanism and requires only a small number of counters to find potential victims, which makes SAP readily implementable on top of existing router mechanisms.

Multiple co-channel WLAN BSSes (i.e., WLAN cells) overlapping in coverage are generally considered undesirable because members of the OBSSes compete for channel access, which typically increases the contention level of wireless medium access and reduces overall system performance. In this paper, we propose to use channel access throttling (CAT) for managing Wireless LAN radio resources for overlapping BSSes (OBSSes). CAT provides an access point (AP) of each BSS with a mechanism to control channel access parameters of its member stations on the fly. By coordinating the CAT operations of the OBSS APs, we can enable privileged channel access to an individual BSS at a particular time, for example, by assigning high priority access parameters to member stations associated with the BSS. By controlling how much each BSS may be given the privileged channel access, we can also achieve a proportional partitioning of channel capacity among OBSSes. We present evaluation results obtained from both simulations and experiments using testbed built with commercial off-the-shelf (COTS) WLAN hardware and open-source device driver. Our results show that with CAT, not only can we proportionally partition channel capacity among the OBSSes, but also improve channel utilization efficiency and increase overall capacity.

A Shrew attack, which uses a low-rate burst carefully designed to exploit TCP's retransmission timeout mechanism, can throttle the bandwidth of a TCP flow in a stealthy manner. While such an attack can significantly degrade the performance of all TCP-based protocols and services including Internet routing (e.g., BGP), no existing scheme clearly solves the problem in real network scenarios. In this paper, we propose a simple protection mechanism, called SAP (Shrew Attack Protection), for defending against a Shrew attack. Rather than attempting to track and isolate Shrew attackers, SAP identifies TCP victims by monitoring their drop rates and preferentially admits those packets from victims with high drop rates to the output queue. This is to ensure that well-behaved TCP sessions can retain their bandwidth shares. Our simulations indicate that under a Shrew attack, SAP can prevent TCP sessions from closing, and effectively enable TCP flows to maintain high throughput. SAP is a destination-port-based mechanism and requires only a small number of counters to find potential victims, which makes SAP readily implementable on top of existing router mechanisms.

Internet Protocol Television (IPTV) has emerged as a new delivery method for TV. In contrast with native broadcast in traditional cable and satellite TV system, video streams in IPTV are encoded in IP packets and distributed using IP unicast and multicast. This new architecture has been strategically embraced by ISPs across the globe, recognizing the opportunity for new services and its potential toward a more interactive style of TV watching experience in the future. Since user activities such as channel switches in IPTV impose workload beyond local TV or set-top box (different from broadcast TV systems), it becomes essential to characterize and model the aggregate user activities in an IPTV network to support various system design and performance evaluation functions such as network capacity planning. In this work, we perform an in-depth study on several intrinsic characteristics of IPTV user activities by analyzing the real data collected from an operational nation-wide IPTV system. We further generalize the findings and develop a series of models for capturing both the probability distribution and time-dynamics of user activities. We then combine theses models to design an IPTV user activity workload generation tool called SIMUL WATCH, which takes a small number of input parameters and generates synthetic workload traces that mimic a set of real users watching IPTV. We validate all the models and the prototype of SIMUL WATCH using the real traces. In particular, we show that SIMUL WATCH can estimate the unicast and multicast traffic accurately, proving itself as a useful tool in driving the performance study in IPTV systems.

Broadcasting is a fundamental operation in wireless networks and plays an important role in the communication protocol design. In multihop wireless networks, however, interference at a node due to simultaneous transmissions from its neighbors makes it non-trivial to design a minimum-latency broadcast algorithm, which is known to be NP-complete. We present a simple 12-approximation algorithm for the one-to-all broadcast problem that improves all previously known guarantees for this problem. We then consider the all-to-all broadcast problem where each node sends its own message to all other nodes. For the all-to-all broadcast problem, we present two algorithms with approximation ratios of 20 and 34, improving the best result available in the literature. Finally, we report experimental evaluation of our algorithms. Our studies indicate that our algorithms perform much better in practice than the worst-case guarantees provided in the theoretical analysis and achieve up to 37% performance improvement over existing schemes.

In IEEE 802.11 Wireless LAN (WLAN) systems, techniques such as acknowledgement, retransmission, and transmission rate adaptation, are frame-level mechanisms designed for combating transmission errors. Recently sub-frame level mechanisms such as frame combining have been proposed by the research community. In this paper, we present results obtained from our bit error study for identifying sub-frame error patterns because we believe that identifiable bit error patterns can potentially introduce new opportunities in channel coding, network coding, forward error correction (FEC), and frame combining mechanisms. We have constructed a number of IEEE 802.11 wireless LAN testbeds and conducted extensive experiments to study the characteristics of bit errors and their location distribution. Conventional wisdom dictates that bit error probability is the result of channel condition and ought to follow corresponding distribution. However our measurement results identify three repeatable bit error patterns that are not induced by channel conditions. We have verified that such error patterns are present in WLAN transmissions in different physical environments and across different wireless LAN hardware platforms. We also discuss our current hypotheses for the reasons behind these bit error probability patterns and how identifying these patterns may help improving WLAN transmission robustness.

The de facto QoS channel access method for the IEEE 802.11 Wireless LANs is the Enhanced Distributed Channel Access (EDCA) mechanism, which differentiates transmission treatments for data frames belonging to different traffic categories with four different levels of channel access priority. In this paper, we propose extending EDCA with Channel Access Throttling (CAT) for more flexible and efficient QoS support. By assigning different member stations different channel access parameters, CAT differentiates channel access priorities not between traffic categories but between member stations. Then by dynamically changing the channel access parameters of each member station based on a pre-computed schedule, CAT enables EDCA WLANs the benefits of scheduled access QoS. We also present evaluation results of CAT obtained from both simulations and experiments conducted using off-the-shelf WLAN hardware and open-source device driver. Our results show that CAT can proportionally partition channel capacity, significantly improve performance of multimedia applications, effectively achieve performance protection for admitted flows, and increase per cell VoIP call capacity by up to 41%.

Anycast-based content delivery networks (CDNs) have many properties that make them ideal for the large scale distribution of content on the Internet. However, because routing changes can result in a change of the endpoint that terminates the TCP session, TCP session disruption remains a concern for anycast CDNs, especially for large file downloads. In this paper we demonstrate that this problem does not require any complex solutions. In particular, we present the design of a simple, yet efficient, mechanism to handle session disruptions due to endpoint changes. With our mechanism, a client can continue the download of the content from the point at which it was before the endpoint change. Furthermore, CDN servers purge the TCP connection state quickly to handle frequent switching with low system overhead.

Because it is an integral part of the Internet routing apparatus, and because it allows multiple instances of the same service to be "naturally" discovered, IP Anycast has many attractive features for any service that involve the replication of multiple instances across the Internet. While briefly considered as an enabler when content distribution networks (CDNs) first emerged, the use of IP Anycast was deemed infeasible in that environment. The main reasons for this decision were the lack of load awareness of IP Anycast and unwanted side effects of Internet routing changes on the IP Anycast mechanism. Prompted by recent developments in route control technology, as well as a better understanding of the behavior of IP Anycast in operational settings, we revisit this decision and propose a load-aware IP Anycast CDN architecture that addresses these concerns while benefiting from inherent IP Anycast features. Our architecture makes use of route control mechanisms to take server and network load into account to realize load-aware Anycast. We show that the resulting redirection requirements can be formulated as a Generalized Assignment Problem and present practical algorithms that address these requirements while at the same time limiting session disruptions that plague regular IP Anycast. We evaluate our algorithms through trace based simulation using traces obtained from an operation CDN network.

We consider the problem of finding "backbones" in multihop wireless networks. The backbone provides end-to-end connectivity, allowing nonbackbone nodes to save energy since they do not have to route nonlocal data or participate in the routing protocol. Ideally, such a backbone would be small, consist primarily of high capacity nodes, and remain connected even when nodes are mobile or fail. Unfortunately, it is often infeasible to construct a backbone that has all of these properties; e.g., a small optimal backbone is often too sparse to handle node failures or high mobility. We present a parameterized backbone construction algorithm that permits explicit trade-offs between backbone size, resilience to node movement and failure, energy consumption, and path lengths. We prove that our scheme can construct essentially best possible backbones (with respect to energy consumption and backbone size) when the network is relatively static. We generalize our scheme to build more robust structures better suited to networks with higher mobility. We present a distributed protocol based upon our algorithm and show that this protocol builds and maintains a connected backbone in dynamic networks. Finally, we present detailed packet-level simulation results to evaluate and compare our scheme with existing energy-saving techniques. Our results show that, depending on the network environment, our scheme increases network lifetimes by 20 percent to 220 percent without adversely affecting delivery ratio or end-to-end latency.

We present a protocol to construct routing backbones in wireless networks composed of selfish participants. Backbones are inherently cooperative, so constructing them in selfish environments is particularly difficult; participants want a backbone to exist (soothers relay their packets) but do not want to join the backbone (so they do not have to relay packets for others).

We consider the channel assignment problem for multihop wireless networks in which nodes have multiple interfaces. Given the number of interfaces at each node and available channels in the system, we find a feasible channel assignment to improve network performance. Even when routing is given, finding a channel assignment for optimal performance is NP-hard. We present the SAFE (skeleton assisted partition FrEe) channel assignment scheme, which uses randomized channel assignment in a distributed manner while maintaining network connectivity. SAFE can utilize all independent channels in the system while attempting to distribute edges sharing a particular channel evenly throughout the network. To handle topology change and incremental deployment better, SAFE decouples the channel assignment problem from routing. Our simulation results show that SAFE significantly improves network performance in terms of throughput and delay and is comparable to the best prior centralized scheme that jointly considers routing and channel assignment

Despite improvements in wireless access technologies such as 3G or 802.11x, ubiquitous data access has remained a challenge, mainly due to the lack of inexpensive, pervasive backhaul connections from access points to the Internet. With the recent WiMAX standard for high-speed, non-line-of-sight fixed wireless links, multihop wireless backhauls might now overcome this bottleneck. However an important remaining challenge is to provide rate and delay guarantees for customer connections similar to wired backhauls. We provide several schemes for performing admission control for connections with QoS requirements over a multihop wireless backhaul. This is the first work to address both rate and delay requirements for connections. Our admission control algorithms first construct appropriate tree-based topologies connecting wireless backhaul nodes to a wired gateway and then admit the best subset of connections while respecting their rate and delay requirements. Alternately, we admit all the connections with appropriate degradation of their QoS requirements

We propose a new link metric called normalized advance (NADV) for geographic routing in multihop wireless networks. NADV selects neighbors with the optimal trade-off between proximity and link cost. Coupled with the local next hop decision in geographic routing, NADV enables an adaptive and efficient cost-aware routing strategy. Depending on the objective or message priority, applications can use the NADV framework to minimize various types of link cost.We present efficient methods for link cost estimation and perform detailed simulations in diverse scenarios. Our results show that NADV outperforms current schemes in many aspects: for example, in high noise environments with frequent packet losses, the use of NADV leads to 81% higher delivery ratio. When compared to centralized routing under certain settings, geographic routing using NADV finds paths whose cost is close to the optimum.

We propose a multi-hop wireless LAN architecture and demonstrate its benefits to wireless clients. For this architecture, we define implementation paths that allow interoperation with existing wireless LANs which can lead to an incremental deployment of this system. We quantify the performance benefits of the proposed schemes through measurements in realistic wireless LAN environments. We also examine the performance of such multi-hop wireless LANs through detailed simulation studies. Our results show that these multi-hop extensions can significantly improve the wireless access experience (in terms of data throughput, latency, etc.) for clients who enable such mechanisms. More interestingly, when multi-hop extensions are enabled by some of the clients, it also positively impacts the performance at other clients that are completely unaware of these extensions.

We present a low-overhead media streaming system, called SRMS (Scalable Resilient Media Streaming) that can be used to scalably deliver streaming data to a large group of receivers. SRMS uses overlay multicast for data distribution. to a large group of users. SRMS leverages a probabilistic loss recovery technique to provide high data delivery guarantees even under large network losses and overlay node failures. The clients in the SRMS system are able to interoperate with existing media streaming servers that use RTP for data transport. One of the interesting features of SRMS is that it can simultaneously support clients with disparate access bandwidths. It enables the necessary bandwidth adaptations using standard Real-time Transport Protocol (RTP) mechanisms, e.g. RTP translators. We have implemented and evaluated the SRMS system in detail on an emulated network as well as on a wide-area testbed with up to 128 clients. Our results show that clients using SRMS achieve high (97%) data delivery ratios with low overheads (<5%) even for a very dynamic network (up to five membership changes per minute).

A distributed scheme for trust inference in peer-to-peer networks is presented. Our work is in context of the NICE system, which is a platform for implementing cooperative applications over the Internet. We describe a technique for efficiently storing user reputation information in a completely decentralized manner, and show how this information can be used to efficiently identify noncooperative users in NICE. We present a simulation based study of our algorithms, in which we show our scheme scales to thousands of users using modest amounts of storage, processing, and bandwidth at any individual node. Lastly, we show that our scheme is robust and can form cooperative groups in systems where the vast majority of users are malicious.

We introduce PRM (Probabilistic Resilient Multicast): a multicast data recovery scheme that improves data delivery ratios while maintaining low end-to-end latencies. PRM has both a proactive and a reactive component; in this paper we describe how PRM can be used to improve the performance of application-layer multicast protocols, especially when there are high packet losses and host failures. Further, using analytic techniques, we show that PRM can guarantee arbitrarily high data delivery ratios and low latency bounds. As a detailed case study, we show how PRM can be applied to the NICE application-layer multicast protocol. We present detailed simulations of the PRM-enhanced NICE protocol for 10,000 node Internet-like topologies. Simulations show that PRM achieves a high delivery ratio (> 97%) with a low latency bound (600 ms) for environments with high end-to-end network losses (1-5%) and high topology change rates (5 changes per second) while incurring very low overheads (< 5%).

An ad hoc network is a multi-hop wireless network formed by a collection of mobile nodes without the intervention of fixed infrastructure. Limited bandwidth and a high degree of mobility require that routing protocols for ad hoc networks be robust, simple, and energy-conserving. This paper proposes a new ad hoc multicast routing protocol called neighbor-supporting multicast protocol (NSMP). NSMP adopts a mesh structure to enhance resilience against mobility. NSMP utilizes node locality to reduce the overhead of route failure recovery and mesh maintenance. NSMP also attempts to improve route efficiency and reduce data transmissions. Our simulation results show that NSMP delivers packets efficiently while substantially reducing control overhead in various environments.