Vasanth Iyer

Min Loading Max Reusability Fusion Classifiers for Sensor Data Model

With the deployment of sensor network applications there is lot of breakthrough in digital or smart sensor designs. These smart sensors have a dedicated processor which allows interfacing to many sensors which measure ambient readings of temperature, pressure and humidity of the environment. These sensors once calibrated they function independently running using a self-powered battery which operates 220-mA hours of operation and smart sensor uses 2.45 uA to calibrate and measure the values giving a lifetime of 220 mAh/2.45 muA=10.25 years. A sensor network typically uses 100 to 1,000 of these sensor nodes to monitor the measured values in a given region of deployment. The modeled network use data forwarding, energy conservation and fault tolerance and some of the deployment use secure communication as these are wireless based sensors. This networking model is optimized to use very less central coordination as it is an infrastructure less protocol. Due to the nature of the sensor model we develop two categories of fusion techniques. One is purely based on distributed quantitative routing which is defined by min node loading and max node reusability fusion technique. This fusion technique initially classifies the robustness of the distributed algorithms during its lifetime by defining them as tamely faulty sensors and widely faulty sensors. This model is further qualitatively extended for decision making defined by min data redundancy and max sensor fault-tolerant using a measurement independent probability data model for each clustering algorithm.

Distributed Source Coding for Sensor Data Model and Estimation of Cluster Head Errors Using Bayesian and K-Near Neighborhood Classifiers in Deployment of Dense Wireless Sensor Networks

The lifetime calculation of large dense sensor networks with fixed energy resources and the remaining residual energy have shown that for a constant energy resource in a sensor network the fault rate at the cluster head is network size invariant when using the network layer with no MAC losses.Even after increasing the battery capacities in the nodes the total lifetime does not increase after a max limit of 8 times. As this is a serious limitation lots of research has been done at the MAC layer which allows to adapt to the specific connectivity, traffic and channel polling needs for sensor networks. There have been lots of MAC protocols which allow to control the channel polling of new radios which are available to sensor nodes to communicate. This further reduces the communication overhead by idling and sleep scheduling thus extending the lifetime of the monitoring application. We address the two issues which effects the distributed characteristics and performance of connected MAC nodes. (1) To determine the theoretical minimum rate based on joint coding for a correlated data source at the singlehop, (2a) to estimate cluster head errors using Bayesian rule for routing using persistence clustering when node densities are the same and stored using prior probability at the network layer, (2b) to estimate the upper bound of routing errors when using passive clustering were the node densities at the multi-hop MACS are unknown and not stored at the multi-hop nodes a priori. In this paper we evaluate many MAC based sensor network protocols and study the effects on sensor network lifetime. A renewable energy MAC routing protocol is designed when the probabilities of active nodes are not known a priori. From theoretical derivations we show that for a Bayesian rule with known class densities of omega1, omega 2 with expected error P* is bounded by max error rate of P = 2P* for single-hop. We study the effects of energy losses using cross-layer simulation of large sensor network MACS setup, the error rate which effect finding sufficient node densities to have reliable multi-hop communications due to unknown node densities. The simulation results show that even though the lifetime is comparable the expected Bayesian posterior probability error bound is close or higher than P ≥ 2P.

INSPIRE-DB: Intelligent Networks Sensor Processing of Information Using Resilient Encoded-Hash DataBase

Sensor networks consist of small motes attached with sensors to measure ambient parameters like temperature, humidity and light. As these motes are unreliable due to wireless link quality and also the data measuring sensors cannot be calibrated accurately for a given applications need. The unique data fusion needs are that parameter being measured is distributed across the network and needs to be computed reliably and with minimum overhead and redundancy due to data value being correlated. We show the asymptotic complexity of topology control when applied to power-aware routing is scalable and argue that the accuracy and reliability of the estimated sensor values can be accurately predicted for the physical value being sensed and aggregating. A prefixbased routing protocol is used for data-centric storage, which allows querying distributed parameters using a KEY, VALUE pairs without the need of the sensor node to know its exact geographic information. Intelligent sensor information processing, which is driven by these requirements, is discussed under the framework INSPIRE-DB.

FARMS: Fusionable ambient renewable MACS

We address two critical issues in wireless sensor networks: (1) an extension of common Quality of Service (QoS) parameters to study the effects of ultra-low duty cycling applications, and (2) propose a new WSN passive clustering routing protocol using sleep cycles based on available renewable energy resources : Fusion Ambient Renewable MACS (FARMS). The results from lifetime based QoS in a time synchronized deployment show that for a best effort QoS multi-hop deployment with varying percentage of cluster heads, the lifetime is network size and protocol invariant. However, low sensing ranges result in dense networks and thus it becomes necessary to achieve an efficient medium-access protocol subject to power constraints. We present cross-layer energy dissipation per node and show the performance of the network by varying duty-cycles. The study of sensor FARMS harvesting applications allows to measure the impact on idle, sleep and renewable energy cycles and their unique deployment (in terms of density) needs as all the sensor are not active at all times. We show that efficiency of sensor deployment QoS can be provided in terms of distributed load balancing (20% static clustering) at each node, power-aware sleep scheduling (2X increase in lifetime), data aggregation efficiency(B-MAC performs 3X times better than CSMA) in multi-hop passive clustering implementations.

Modeling Unreliable Data and Sensors: Using F-measure Attribute Performance with Test Samples from Low-Cost Sensors

Building a high performance classifier requires training with labeled data, which is supervised and allows generalizing the classifiers decision boundary and in practice most of the data is unlabeled, newer algorithms needs to be learn by knowledge discovery. Sufficient training data are collected in the form of empirical evidence, which have labeled positive and negative samples to build the hypothesis. The hypothesis is constructed by the conjunction of the attributes, which can be learnt by machine learning algorithm. In this paper, we work with two forms of ranking weights, precision and relevance, which help in finding hidden patterns and prediction future events. Empirical evidence for a weather patterns and tracking of a phenomenon needs to accurately extract the attributes and label the training samples, which is a very laborious and time-consuming effort. Automating weather prediction algorithms, which are trained by supervised learning, needs to be generalized so that it can be tested with unreliable and noisy weather data from low cost sensors. We use a training data from previous forest fires events, the datasets containing all the attributes are labeled using manual data logs for a given geographical area. The labeled original dataset is mapped to the data collected from on-line sensors, which further improves the accuracy of the training set. As some of classes have very few samples, which are related to the peak fire seasons, domain specific knowledge are added by sensor measurements and Fire Weather Index (FWI) to help accurately model the events. We show that training accuracy of the small forest fire classifier using attributes from manual logs is enhanced by 30% by using sensor data. The rare and hard to classify large forest fires are 95% accurately classified by using the new Fire Weather Index (FWI). We also show that our framework is more robust to outliers from noisy sensor measurements by accounting for in the model parameters. The model allows further generalization for linearly and non-linearly separable datasets by estimating the parameters d and minimum allowable error ? for hypothesis, sampling accuracy and cross validation.

Effects of channel SNR in mobile cognitive radios and coexisting deployment of cognitive wireless sensor networks

In this paper, we describe the cognitive radios sharing the spectrum with licensed users and its effects on operational coexistence with unlicensed users. Due to the unlicensed spectrum band growing needs and usage by many IEEE 802.11 protocols, normal wireless radio operation sees high interference leading to high error rates on operational environments. We study the licensed bands and the characteristics of the unlicensed bands in general and more specific to radio characterization of individual radios and cognitive deployment of sensor networks and its effect on lifetime. The cognitive radio signals detection algorithm for this probabilistic model for the unlicensed users, uses a mobility model which takes into account the threshold variable ratio Eb/No and also calculates the lower-bound of the combined value of secondary user interference for overlapping frequencies with the primary user. By using simulation, we detect the primary user when the radio frequencies are known a priori and compare it when the frequencies are unknown. In our analysis we exploit the similarity measure seen at each sub-channel frequency, which are due to multiple paths of the same reflected signal by maximizing the correlated information of the correlation matrix. For the general case the co-variance matrix for blind source separation, we use ICA de-correlation methods and show that cognitive radio can efficiently identify users in complex situations. The effects of large deployment and cognitive sensor network are studied for a family of 802.15.4 radios adapting to power-aware algorithms.

SPOTLESS: Similarity patterns of trajectories in label-less sensor streams

The process of inversion, estimation and reconstruction of the sensor quality matrix, allows modeling the precision and accuracy, and in general the reliability of the model. When the sensor data ranges are not known a priori, current systems do not train on new data samples, rather they approximate based on the parameters global average value, losing most of the spatial and temporal features. The proposed model, which we call SPOTLESS, checks the spatial integrity and temporal plausibility of streams generated by mobility patterns due to varying channel conditions. We define a minimum quality of the measured sensor data as local stream (QoD) requirements to give high precision by using distributed labeled training. In our SPOTLESS datacleaning steps, to account for packet errors due to varying channel conditions, a soft-phy based decoding is selected for various Bit Error Rates (BER), minimizing packet loss at the mobile receiver. Numerical experiments for Rayleigh fading channels and mobile BER model examples are compared with large deployment of ground sensor collecting static data streams and Data MULE collecting multi-hop temporal data from the sensor to provide hypothetical parameter accuracy. Our results were obtained in the context of provisioning a minimum precision and accuracy stream (QoD) required for 802.15.4 mobile services. SPOTLESS data-cleaning algorithm coding provides 90% precision for static streams, and increases the plausible relevance of multi-hop mobile streams by 85% for task-based learning.

Level controlled clustering in wireless sensor networks

Sensor networks can be used for various applications areas like health, military, environmental etc. For different application areas, there are different technical issues that researchers are currently resolving. This paper deals with a distributed sensor network employing level controlled clustering. Level controlled clustering is a technique that uses leveling and clustering together. This technique reduces the number of messages in the direction of base station and thereby increases the life time of the wireless sensor network. We divide the network into levels of different power levels. By using various power levels at base station, the sensor field is hierarchically partitioned into levels of increasing radius each level containing various sensor nodes. Leveling divides the network into logical zones based on proximity from base station, whereby the packet is transmitted from a node in the next zone with lesser depth. The transmission probability is fixed. The primary advantage of this protocol is transmitting a critical event and at the same time conserving life time of the network for future monitoring.

C-ERROR simulator for development for sensor and location aware sensing applications

Distributed wireless sensor applications are useful for visualizing spatially and geographically related data such as location, neighborhood, weather, and measuring specific changes in the environment. Desires to augment these interfaces with additional specifications needed for distributed applications such as Power-Aware, Fault-tolerance and Processor agnostic deployment requirements have led to create a custom distributed Network Embedded Test-Bed that locally aggregate the measured signal from individual sensors and send it to a central coordinator for combined processing. We envision publishing and querying real-time (e.g. from sensors) over such distributed sensor farm applications which are deployed wirelessly and form a large sensor network. Existing solutions, although useful for writing the simple applications mentioned above, have several drawbacks in achieving this vision. First, publishing even a single stream of data as a useful service is a non-trivial task. Much useful data is not being stored yet because the need for managing a sensor farm has lots of complexities which make them unreliable in terms of polling time and communications costs. Second, existing applications are mutually incompatible and are processor centric and needs many ports which may introduce un-reliability. Third communication costs are not scalable to handle a sensor farm application and it does not provide an easy way to extend such a Network Embedded Test-Bed. The Network Embedded Test-Bed project aims to address these challenges, we like to model existing applications needs into a cross layer sensor network simulator called C-ERROR(Cross Layer Reusable Resource Optimized Routing) which allows different clustering algorithms to be integrated and measure its performance at each layer of the stack. To have a platform independent sensor OS and a scheduler which allows creating sensing tasks that have real-time constraints.

Entropy based variable rate compression for low-bandwidth multi-media streams

Trans-coding of Audio and video content is a recompression based on pre-processing the input and converting into a stream able wireless format over low bandwidth network which does not maintain an exactly constant bit rate. As the portable devices have an circular buffer other ad hoc networks which handles the incoming stream the encoder time indexes the stream bits according to the first frames request and compresses and delivers packet according to the dynamic channel bandwidth. The circular buffer is said to contain the start sequence of the frame and the remaining are delivered before the Presentation Time field without ever overrunning the buffer. The compression model uses a similar algorithm as Huffman codes and the streams are optimized to be able easily seek into each frame by using a variable unitary part and fixed offset which is multiple of 2 for easy decoding. Only the frames are transmitted and the values are shared using a lookup table between the encoder and the decoder. In this paper the encoder uses multi-channel audio to mix different PCM samples across 64 volume levels of dynamic range to achieve trans-coding fps. In this paper we define an on-demand adaptive algorithm called trans-repairing for constrained devices which takes original 2n symbols and re-pairs it into m n symbols which are the current needs of low-power, low-bandwidth heterogeneous GPRS, ZigBee and other ad-hoc networks forming a collaborative multi-media smart wireless network.