Arsalan Tavakoli

An empirical study of low-power wireless

We present empirical measurements of the packet delivery performance of the latest sensor platforms: Micaz and Telos motes. In this article, we present observations that have implications to a set of common assumptions protocol designers make while designing sensornet protocols—specifically—the MAC and network layer protocols. We first distill these common assumptions in to a conceptual model and show how our observations support or dispute these assumptions. We also present case studies of protocols that do not make these assumptions. Understanding the implications of these observations to the conceptual model can improve future protocol designs.

Flush: a reliable bulk transport protocol for multihop wireless networks

We present Flush, a reliable, high goodput bulk data transport protocol for wireless sensor networks. Flush provides end-to-end reliability, reduces transfer time, and adapts to time-varying network conditions. It achieves these properties using end-to-end acknowledgments, implicit snooping of control information, and a rate-control algorithm that operates at each hop along a flow. Using several real network topologies, we show that Flush closely tracks or exceeds the maximum goodput achievable by a hand-tuned but fixed rate for each hop over a wide range of path lengths and varying network conditions. Flush is scalable; its effective bandwidth over a 48-hop wireless network is approximately one-third of the rate achievable over one hop. The design of Flush is simplified by assuming that different flows do not interfere with each other, a reasonable restriction for many sensornet applications that collect bulk data in a coordinated fashion, like structural health monitoring, volcanic activity monitoring, or protocol evaluation. We collected all of the performance data presented in this paper using Flush itself.

A policy-aware switching layer for data centers

Data centers deploy a variety of middleboxes (e.g., firewalls, load balancers and SSL offloaders) to protect, manage and improve the performance of applications and services they run. Since existing networks provide limited support for middleboxes, administrators typically overload path selection mechanisms to coerce traffic through the desired sequences of middleboxes placed on the network path. These ad-hoc practices result in a data center network that is hard to configure and maintain, wastes middlebox resources, and cannot guarantee middlebox traversal under network churn. To address these issues, we propose the policy-aware switching layer or PLayer, a new layer-2 for data centers consisting of inter-connected policy-aware switches or pswitches. Unmodified middleboxes are placed off the network path by plugging them into pswitches. Based on policies specified by administrators, pswitches explicitly forward different types of traffic through different sequences of middleboxes. Experiments using our prototype software pswitches suggest that the PLayer is flexible, uses middleboxes efficiently, and guarantees correct middlebox traversal under churn.

The design and implementation of a declarative sensor network system

Sensor networks are notoriously difficult to program, given that they encompass the complexities of both distributed and embedded systems. To address this problem, we present the design and implementation of a declarative sensor network platform, DSN: a declarative language, compiler and runtime suitable for programming a broad range of sensornet applications. We demonstrate that our approach is a natural fit for sensor networks by specifying several very different classes of traditional sensor network protocols, services and applications entirely declaratively -- these include tree and geographic routing, link estimation, data collection, event tracking, version coherency, and localization. To our knowledge, this is the first time these disparate sensornet tasks have been addressed by a single high-level programming environment. Moreover, the declarative approach accommodates the desire for architectural flexibility and simple management of limited resources. Our results suggest that the declarative approach is well-suited to sensor networks, and that it can produce concise and flexible code by focusing on what the code is doing, and not on how it is doing it.

Hydro: A Hybrid Routing Protocol for Low-Power and Lossy Networks

Existing routing protocols for sensor networks ei- ther exclusively focus on collection-based traffic, or optimize for point-to-point traffic in a homogeneous network. As these networks become more general, a mix of these workloads in a heterogeneous setting is expected, while still abiding by the resource constraints of low- power and lossy networks (L2Ns). Our design leverages the predominantly two-tiered topology of L2N deployments, with capable border routers supplementing resource-starved in- network nodes, and optimizes for the typical traffic workloads consisting mainly of collection based traffic with specific instances of point-to-point traffic. We present Hydro, a hybrid routing protocol that combines local agility with centralized control. In-network nodes use distributed DAG formation to provide default routes to border routers, concurrently forming the foundation for triangle point- to-point routing. Border Routers build a global, but typically incomplete, view of the network using topology reports received from in- network nodes, and subsequently install optimized routes in the network for active point-to-point flows. Building on the vast existing literature on distributed DAG for- mation in L2Ns and centralized routing in large-scale networks, our contribution lies in the merging of these ideas in the form of a routing protocol that addresses the needs of L2Ns while remaining grounded in their inherent constraints. Evaluations across testbeds and deployments demonstrate the performance and functionality of Hydro across a variety of workloads and network conditions.

An architecture for energy management in wireless sensor networks

Sensornets are becoming more widely adopted for commercial and scientific use and, in settings where battery replacement or recharging is difficult, it is important that sensornets have long and predictable lifetimes. We thus expect energy management to play an increasingly important role in meeting user requirements. Today, system developers seek a balance between network lifetime and performance, but recent history shows that unexpected and dynamic environmental conditions often scuttle expected energy budgets. For example, many nodes in the Great Duck Island deployment were conjectured to have died prematurely because unexpected overhearing of traffic caused radios to become operational for longer than originally predicted [22]. This pattern was repeated in the Redwoods deployment, but for a supposedly different reason: some nodes seemingly died prematurely because they became disconnected from the wireless network and depleted their batteries trying to reconnect [24]. Even systems augmented with energy harvesting are still susceptible to this type of problem. In the Trio Testbed, seasonal and daily variations in solar power, the angle of inclination of the solar cell, the effect of dirt and bird droppings on the cells, and the inefficiencies in power storage and transfer resulted in node duty cycles ranging from 20% to 100% [5]. The issues with these deployments arise from mistaken assumptions, unforeseen difficulties, and unpredictable environmental dynamics. Solutions to these issues take two extreme approaches. At one extreme, some have proposed runtime adaptation to meet lifetime requirements [16] or energy availability [11, 10]. While promising, these efforts have addressed rather coarse-grained, high-level adaptation – for example, by adjusting sampling rates or varying the system-wide duty cycle – but they remain silent on prioritizing energy usage in a fine-grained and flexible manner. At the other extreme, low-level energy management mechanisms that give direct control over the hardware to multiple entities (e.g. network protocols) can be tedious to implement and difficult to debug because of the lack of any isolation. Arbitration can address the isolation problem, but it does not enable runtime adaptation to varying workloads [12]. We believe that using an energy management architecture would alleviate or even prevent these types of problems. SecEnforcement

On-line sensing task optimization for shared sensors

Shared sensing infrastructures that allow multiple applications to share deployed sensors are emerging and Internet protocol based access for such sensors has already been prototyped and deployed. As a large number of applications start accessing shared sensors, the efficiency of resource usage at the embedded nodes and in the network infrastructure supporting them becomes a concern. To address this, we develop methods that detect when common data and common stream processing is requested by multiple applications, including cases where only some of the data is shared or only intermediate processing steps are common. The communication and processing is then modified to eliminate the redundancies. Specifically, we use an interval-cover graph to minimize communication redundancies and a joint data flow graph optimization to remove computational redundancies. Both methods operate online and allow application requests to be dynamically added or removed. The proposed methods are evaluated using applications on a road traffic sensing infrastructure.

A declarative sensornet architecture

Increased code reuse, independent development, and interoperability are three key missing characteristics of sensornet programming today that motivate the need for an overall sensornet architecture. Architecture traditionally implies a componentized modular framework in which narrow interfaces provide the only form of communication between layers encapsulating specific functions and services. The Internet architecture provides the classic example of the modular approach.

A modular sensornet architecture: past, present, and future directions

ion that supports existing implementations, more so than building a narrow waist that exports a set of services that new implementations are expected to adhere to. While future applications may use such features, the disregard for these interfaces by current protocols, coupled with the desire to maintain a lean narrow waist, leads to a top-down focus. In designing the unifying link layer abstraction for T2 [20], our approach has been quite different. We began with a foundation similar to the previous version, pruning unneeded services and functionality. More importantly we synthesized a list of requirements for a diverse set of protocols and applications in order to ensure support in the upcoming version. We also focused on integrating cross-layer services, such as security and power management, working closely with developers of these stand-alone components and focused architectures. The result is a new link abstraction which we feel is lean, yet provides the essential set of services needed to support the majority of higher-level services.