Sanjib Sur

60 GHz Indoor Networking through Flexible Beams: A Link-Level Profiling

60 GHz technology holds tremendous potential to upgrade wireless link throughput to Gbps level. To overcome inherent vulnerability to attenuation, 60 GHz radios communicate by forming highly-directional electronically-steerable beams. Standards like IEEE 802.11ad have tailored MAC/PHY protocols to such flexible-beam 60 GHz networks. However, lack of a reconfigurable platform has thwarted a realistic proof-of-concept evaluation. In this paper, we conduct an in-depth measurement of indoor 60 GHz networks using a first-of-its-kind software-radio platform. Our measurement focuses on the link-level behavior with three major perspectives: (i) coverage and bit-rate of a single link, and implications for 60 GHz MIMO; (ii) impact of beam-steering on network performance, particularly under human blockage and device mobility; (iii) spatial reuse between flexible beams. Our study dispels some common myths, and reveals key challenges in maintaining robust flexible-beam connection. We propose new principles that can tackle such challenges based on unique properties of 60 GHz channel and cognitive capability of 60 GHz links.

WiFi-Assisted 60 GHz Wireless Networks

Despite years of innovative research and development, gigabit-speed 60 GHz wireless networks are still not mainstream. The main concern for network operators and vendors is the unfavorable propagation characteristics due to short wavelength and high directionality, which renders the 60 GHz links highly vulnerable to blockage and mobility. However, the advent of multi-band chipsets opens the possibility of leveraging the more robust WiFi technology to assist 60 GHz in order to provide seamless, Gbps connectivity. In this paper, we design and implement MUST, an IEEE 802.11-compliant system that provides seamless, high-speed connectivity over multi-band 60 GHz and WiFi devices. MUST has two key design components: (1) a WiFi-assisted 60 GHz link adaptation algorithm, which can instantaneously predict the best beam and PHY rate setting, with zero probing overhead; and (2) a proactive blockage detection and switching algorithm which can re-direct ongoing user traffic to the robust interface within sub-10 ms latency. Our experiments with off-the-shelf 802.11 hardware show that MUST can achieve 25-60% throughput gain over state-of-the-art solutions, while bringing almost 2 orders of magnitude cross-band switching latency improvement.

Autodirective audio capturing through a synchronized smartphone array

High-quality, speaker-location-aware audio capturing has traditionally been realized using dedicated microphone arrays. But high cost and lack of portability prevents such systems from being widely adopted. Todays smartphones are relatively more convenient for audio recording, but the audio quality is much lower in noisy environment and speaker location cannot be readily obtained. In this paper, we design and implement Dia, which leverages smartphone cooperation to overcome the above limitations. Dia supports spontaneous setup, by allowing a group of users to rapidly assemble an array of smartphones to emulate a dedicated microphone array. It employs a novel framework to accurately synchronize the audio I/O clocks of the smartphones. The synchronized smartphone array further enables autodirective audio capturing, i.e., tracking the speakers location, and beamforming the audio capturing towards the speaker to improve audio quality. We implement Dia on a testbed consisting of 8 Android phones. Our experiments demonstrate that Dia can synchronize the microphones of different smartphones with sample-level accuracy. It achieves high localization accuracy, and similar beamforming performance compared with a microphone array with perfect synchronization.

Handwritten Bangla character recognition in machine-printed forms using gradient information and Haar wavelet

A robust and efficient algorithm to recognize handwritten Bangla (Bengali) characters in machine-printed forms is proposed. It is based on the combination of gradient features and Haar wavelet coefficients. The gradient feature is used to capture local characteristics, and for its sensitivity to the usual deformation and idiosyncrasy of handwritten characters, wavelet transform is used for multi-resolution analysis of character images. Such a strategy with combined features captures adequate global characteristics in different scales. Two feature-combination schemes are devised and tested on test images of 4372 instances of 49 characters and 10 numerals, after being trained by a set of 59×25 = 1475 images. Finally, a k-NN classifier is used for the character recognition, which shows 87.65% and 88.95% recognition accuracies for the two schemes.

Practical MU-MIMO user selection on 802.11ac commodity networks

Multi-User MIMO, the hallmark of IEEE 802.11ac and the upcoming 802.11ax, promises significant throughput gains by supporting multiple concurrent data streams to a group of users. However, identifying the best-throughput MU-MIMO groups in commodity 802.11ac networks poses three major challenges: a) Commodity 802.11ac users do not provide full CSI feedback, which has been widely used for MU-MIMO grouping. b) Heterogeneous channel bandwidth users limit grouping opportunities. c) Limited-resource on APs cannot support computationally and memory expensive operations, required by existing algorithms. Hence, state-of-the-art designs are either not portable in 802.11ac APs, or perform poorly, as shown by our testbed experiments. In this paper, we design and implement MUSE, a lightweight user grouping algorithm, which addresses the above challenges. Our experiments with commodity 802.11ac testbeds show MUSE can achieve high throughput gains over existing designs.

Ensuring basic security and preventing replay attack in a query processing application domain in WSN

Nodes in a wireless sensor network are susceptible to various attacks primarily due to their nature of deployment. Therefore, providing security to the network becomes a big challenge. We propose a scheme considering cluster architecture based on LEACH protocol to build a security mechanism in a query-processing paradigm within wireless sensor network. The scheme is capable of thwarting replay attack while ensuring essential properties of security such as authentication, data integrity and data freshness. Our scheme is lightweight as it employs symmetric key cryptography with very short-length key. We simulate our scheme to show its efficacy of providing basic security to the network as well as detecting replay attack in the sensor network. Further we compare our scheme with one of the existing schemes taking packet loss and packet rejection ratio as performance metrics.

Bridging link power asymmetry in mobile whitespace networks

We explore the use of TV White Space (TVWS) wireless networks for providing robust and long range connectivity to vehicles. A key distinctive requirement of TVWS networks is the power asymmetry - the static APs are allowed to transmit at up to 4 W, while the mobile clients in vehicles are limited to only 100 mW. Our measurements reveal that the power asymmetry not only causes severe uplink blackouts but also poses significant coexistence problems, as high-power fixed nodes can easily starve the low-power mobile ones due to carrier sensing loss. To tackle these unique challenges, we propose a cross-layer design of a Direct-Sequence Spread Spectrum (DSSS) based system. We employ an adaptive DSSS mechanism that strategically configures the spreading code, so as to boost uplink coverage while maximizing throughput. We further design a traffic-aware code assignment algorithm for uplink packets to balance the requirement of throughput-intensive and latency-sensitive flows. We have implemented the design on a TVWS software-radio platform on a moving vehicle in an urban environment, and demonstrated that link asymmetry can be completely removed to support realistic application traffic, while the carrier sense loss rate at fixed nodes can be reduced by around 85%.

μ Sec: A Security Protocol for Unicast Communication in Wireless Sensor Networks

Secure communication in wireless sensor networks (WSNs) not only needs to provide the basic security but also to defend various attacks. The challenge in providing security in this network is that the securing mechanism must be lightweight to make it implementable in resource-constrained nodes. In this paper we have devised a link layer protocol for securing unicast communication in wireless sensor network (WSN). The protocol (μSec) is developed in TinyOS platform which is an event-driven operating system used in WSN for networked applications. Our protocol supports the basic security features such as confidentiality, authentication and integrity along with defense against replay attacks. We have modified an existing cryptographic algorithm with a target to minimize computational overhead to make it implementable in WSN. A simple, counter based defense mechanism is proposed to thwart replay attack. Both qualitative and quantitative analyses are performed to measure the efficacy of the protocol. The protocol is compared with some of important security protocols developed around TinyOS. We claim that that μSec, in addition to basic security, thwarts replay attack with same overhead as in other protocols which have considered basic securities only. We further claim that the μSec requires 10% (avg.) less overhead compared to its competitor which also defends replay attack.

Bringing multi-antenna gain to energy-constrained wireless devices

Leveraging the redundancy and parallelism from multiple RF chains, MIMO technology can easily scale wireless link capacity. However, the high power consumption and circuit-area cost prevents MIMO from being adopted by energy-constrained wireless devices. In this paper, we propose Halma, that can boost link capacity using multiple antennas but a single RF chain, thereby, consuming the same power as SISO. While modulating its normal data symbols, a Halma transmitter hops between multiple passive antennas on a per-symbol basis. The antenna hopping pattern implicitly carriers extra data, which the receiver can decode by extracting the index of the active antenna using its channel pattern as a signature. We design Halma by intercepting the antenna switching and channel estimation modules in modern wireless systems, including ZigBee and WiFi. Further, we design a model-driven antenna hopping protocol to balance a tradeoff between link quality and dissimilarity of channel signatures. Remarkably, by leveraging the inherent packet structure in ZigBee, Halmas link capacity can scale well with the number of antennas. Using the WARP software radio, we have implemented Halma along with a ZigBee- and WiFi-based PHY layer. Our experiments demonstrate that Halma can improve ZigBees throughput and energy efficiency by multiple folds under realistic network settings. For WiFi, it consumes similar power as SISO, but boosts throughput across a wide range of link conditions and modulation levels.

Poster: Scoping Environment to Assist 60 GHz Link Deployment

Line-of-Sight blockage by human body is a severe challenge to enable robust 60 GHz directional links. Beamsteering is one feasible solution to overcome this problem by electronically steering phased-array beam towards Non-Line-of-Sight. However, effectiveness of beamsteering depends on the link deployment and a lack of assessment of steering effectiveness may render the link completely blacked-out during human blockage. In this poster, we propose a new technique called BeamScope, that predicts best possible location for a randomly deployed link in an indoor environment without the need of any explicit war-driving. BeamScope first characterizes the environment exploiting measurement from the randomly deployed reference location and then predicts the performance in unobserved locations to suggest a possible re-deployment. The environment characterization is captured through a novel metric and prediction is achieved via how this metric is shared between the reference location and unobserved locations. Our preliminary results show promising accuracy of identifying the best possible alternate location for 60 GHz link to achieve a robust connection during human blockage.