Underwater Acoustic Communications - Seminar

Underwater Acoustic Communications
In the last two decades, underwater acoustic communications has experienced significant progress. The traditional approach for ocean-bottom or ocean-column monitoring is to deploy oceanographic sensors, record the data, and recover the instruments. But this approach failed in real-time monitoring. The ideal solution for real-time monitoring of selected ocean areas for long periods of time is to connect various instruments through wireless links within a network structure. Basic underwater acoustic networks are formed by establishing bidirectional acoustic communication between nodes such as autonomous underwater vehicles (AUVs) and fixed sensors. The network is then connected to a surface station, which can further be connected to terrestrial networks such as the Internet. 

In the last two decades, underwater acoustic communications has experienced significant progress. Communication systems with increased bit-rate and reliability now enable real-time point-to-point links between underwater nodes such as ocean bottom sensors and autonomous underwater vehicles (AUVs). Current researches focused on combining various point-to-point links within a network structure to meet the emerging demand for applications such as environmental data collection, offshore exploration, pollution monitoring, and military surveillance. The traditional approach for ocean-bottom or ocean-column monitoring is to deploy oceanographic sensors, record the data, and recover the instruments. This approach has several disadvantages:

  • The recorded data cannot be recovered until the end of the mission, which can be several months.
  • There is no interactive communication between the underwater instrument and the onshore user. Therefore, it is not possible to reconfigure the system as interesting events occur.
  • If a failure occurs before recovery, data acquisition may stop or all the data may be lost.

The ideal solution for real-time monitoring of selected ocean areas for long periods of time is to connect various instruments through wireless links within a network structure. Basic underwater acoustic networks are formed by establishing bidirectional acoustic communication between nodes such as autonomous underwater vehicles (AUVs) and fixed sensors. The network is then connected to a surface station, which can further be connected to terrestrial networks such as the Internet, through an Ri link. Onshore users can extract real-time data from multiple underwater instruments. After evaluating the obtained data, they can send control messages to individual instruments. Since data is not stored in the underwater instruments, data loss is prevented as long as isolated node failures can be circumvented by reconfiguring the network. A major constraint of underwater acoustic (UWA) networks is the limited energy supply. Whereas the batteries of a wireless modem can easily be replaced on land-based systems, the replacement of an underwater modem battery involves ship time and the retrieval of the modem from the ocean bottom which is costly and time consuming. Therefore, transmission energy is precious in underwater applications. Network protocols should conserve energy by reducing the number of retransmissions, powering down between transactions, and minimizing the energy required for transmission.

Some underwater applications require the network to be deployed quickly without substantial planning, such as in rescue and recovery missions. Therefore the network should be able to determine the node locations and configure itself automatically to provide an efficient data communication environment. Also, if the channel conditions change or some of the nodes fail during the mission, the network should be capable of reconfiguring itself dynamically to continue its operation.

Unlike digital communications through radio channels where data are transmitted by means of electromagnetic waves, acoustic waves are primarily used in underwater channels. The propagation speed of acoustic waves in UWA channels is five orders of magnitude less than that of radio waves. This low propagation speed increases the latency of a packet network. If high latency is overlooked in the design of network protocols for UWA applications, it can reduce the throughput of a network considerably. The available bandwidth of a UWA channel depends critically on transmission loss which increases with both range and frequency and severely limits the available bandwidth. Within this limited bandwidth, the acoustic signals are subjected to time varying multi path which may result in severe inter symbol interference (ISI) and large Doppler shills and spreads, relative to radio channels especially in shallow water channels. Multi path propagation and Doppler effects degrade acoustic signals and limit data throughput. Special processing techniques are needed to combat these channel impairments. Until the beginning of the last decade, due to the channel characteristics of UW channels, modem development was focused on employing non-coherent frequency shift keying (FSK) signals for achieving reliable communication. Although non- coherent FSK systems are effective in UWA channels, their low bandwidth makes them inappropriate for high-data-rate applications such as multi-user networks. The need high-throughput long-range systems have resulted in a focus towards coherent modulation techniques. Today, with the availability of powerful digital signal processing devices, we are able to employ fully coherent phase shift keying (PSK) modulation in underwater communications. A summary of acoustic models is listed in table below:

As the data rate and the range of the systems increase, the complexity of the algorithms grows beyond the capacity of current DSP hardware. Current research is focused on DSP algorithms with decreased complexity and multi-user modems that can operate in a network environment. 

The new Acoustic Underwater Modems bring wireless underwater communication to a new level of performance. The new modems offer significant improvement in performance and reliability over the conventional technology. The modems are based on the patented S2C-Technology opening a new dimension of underwater communication. New and enhanced DSP modulation/demodulation, in combination with new standards in hard- and software design, are the basis for significant improvements, including increased data reliability, extremely higher data rates and reduced system size and weight.

Two types of applications have guided the evolution of underwater networks. One is gathering of environmental data, and the other is surveillance of an underwater area. Typically, the network consists of several types of sensors, some of which are mounted on fixed moorings; the others are mounted on freely moving vehicles. This type of network is called an Autonomous Ocean Sampling Network (AOSN) where the word sampling implies collecting the samples of oceanographic parameters such as temperature, salinity, and underwater currents. For surveillance application, the network consists of a large number of sensors typically bottom-mounted or on slowly crawling robots, that can be quickly deployed, and whose task is to map a shallow water area. An example of such a network is Seaweb.

Network Topologies
There are three basic topologies that can be used to interconnect network nodes which are centralized, distributed and multihop topology. In a centralized network a communication between nodes takes place through a central station, which is sometimes called the hub of the network. The network is connected to a backbone at this central station. This configuration is suitable for deep water networks where surface buoy with both an acoustic and an RF modem acts as the hub and controls the communication to and from ocean bottom instruments. A major disadvantage of this configuration is the presence of a single failure point. If the hub fails, the entire network shuts down. Also due to the limited range of a single modem, the network cannot cover large areas.
The next two topologies belong to peer-to-peer networks. A fully connected peer-to-peer topology provides point-to-point links between every node of the network. Such a topology eliminates the need for routing. However, the output power needed for communicating with widely separated nodes is excessive. Also, a node that is trying to send packets to a far-end node can over power and interfere with the communication between neighboring nodes which is called the near-far problem.
Multihop peer-to-peer networks are formed by establishing communication links between neighboring nodes. Messages are transferred source to destination hopping packets from node to node. Routing of the messages is handled by intelligent algorithms that can adapt to change in conditions. Multihop networks can cover relatively larger areas since the range of the network is determined by the number of nodes rather than the modem range. One of the UWA network design goals is to minimize the energy consumption while providing reliable connectivity between the nodes in the network and the backbone. The network topology is an important parameter that determines energy consumption. The strategy that minimizes energy consumption is multihop peer-to-peer topology. The price paid for the decrease in energy consumption is the need for a sophisticated communication protocol and an increase in packet delay. Therefore, special attention should be given to applications that are sensitive to delays.

Multiple Access Methods
In many information networks, including UWA networks, communication is busty and the amount of time that a user spends transmitting over the channel is usually smaller than the amount of time it stays idle. Thus, network users should share the available frequency and time in an efficient manner by means of a multiple access method. Frequency-division multiple access (FDMA), divides the available frequency bands into sub-bands and assigns each sub-band to an individual user. Due to the severe bandwidth limitations and vulnerability of narrow band systems to fading FDMA systems do not provide an efficient solution for UWA applications.
Instead of dividing a frequency band, time-division multiple access (TDMA) divides a time interval, called a frame, into time-slots. Collision of packets from adjacent time- slots is prevented by including guard times that are proportional to the propagation-delay present in a channel. TDMA systems require very precise synchronization for proper utilization of the time-slots. High latency present in UWA channels require long guard times that limits the efficiency of TDMA. Also, establishing a common timing reference is a difficult task.

Code-division multiple access (CDMA) allows multiple users to transmit simultaneously over the entire frequency band. Signals from different users are distinguished by means of  pseudo-noise (PN) codes that are used for spreading the user messages. The large bandwidth of CDMA channels not only provide resistance to frequency selective fading, but may also take advantage of the time diversity present in the UWA channel by employing rake filters at the receiver in the case of direct sequence CDMA. Spread spectrum signals can he used for resolving collision at the receiver by using multi-user detectors in this way the number of retransmissions and energy requirements of the system are reduced. This property both reduces battery consumption and increases the throughput of the network. Hence CDMA appears to be the most suitable multiple access technique for shallow water acoustic networks.

Routing Algorithms
There are two basic methods used for routing packets through an information network. They are virtual circuit routing and datagram switching.  In virtual circuit routing all the packets of a transaction follow the same path through the network and datagram routing where packets are allowed to pass through different paths. Networks using virtual circuits decide on the path of the communication at the beginning of the transaction.  In datagram switching each node that is involved in a transaction makes a routing decision, which is to determine the next hop of the packet.

Many of the routing methods are based on the shortest path algorithm. In this method each link in the network is assigned a cost which is a function of the physical distance and the level of congestion. The routing algorithm tries to find the shortest path (i.e. the path with the lowest cost) from a source node to a destination node. In the distributed implementation, each node determines the cost of sending a data packet to its neighbors and shares this information with the other nodes of the network. In this way every node maintains a database that reflects the cost of pos routes.  For routing let us consider the most general problem where network nodes are allowed to move. This situation can be viewed as an underwater network with both fixed ocean bottom sensors and AUVs. The instruments temporarily form a network without the aid of any pre-existing infrastructure.

In ad hoc networks, the main problem is to obtain the most recent state of each individual link in the network to decide on the best route for a packet. However if the communication medium is highly variable as in the shallow water acoustic channel the number of routing updates can be very high. Current research on routing focuses on reducing the overhead added by routing messages while at the same time finding the best path which are two conflicting requirements. Also, the effect of long propagation delays and channel asymmetries caused by power control arc issues that need to be addressed when designing network routing protocols to UWA channels.    

Media Access Control Protocols
There are various media access control (MAC) protocols that can be employed to avoid information loss in UWA networks due to packet collisions. We shall focus on the MACA protocol and a variation of this protocol.

The MACA protocol, proposed by Karn used two signaling packets called Request- to-Send (RTS) and Clear-to-Send (CTS). When A wants to send a message to B it first issues an RTS command. If B receives the RTS it sends back a CTS command. As soon as A receives CTS it begins transmission of the data packet. The nodes can probe the channel during the RTS-CTS exchange. The channel state information can be used to set the physical layer parameters such as output power and modulation type. These properties of the MACA protocol are essential for efficient UWA network design. It provides information for reliable communication with minimum energy consumption and has the ability to avoid collisions before they occur. RTS-CTS exchange adds overhead, but the reduction of retransmissions can compensate for this, increase. 

The MACA protocol ensures the reliability of the end-to-end link with the network layer. If some packets of a message are lost due to errors, the final destination node will ask the originating source to retransmit the lost packets. On highly reliable links this approach increases throughput, since it eliminates the need to send individual acknowledgments for each hop. In case of poor-quality communication channels, a message will most probably contain erroneous packets. Recovering the errors in a data packet at the network layer will require excessive delay. Generally, error correction is better performed at the data link layer for channels of low reliability such as radio or shallow water acoustic channels.

The performance and reliability of the MACA protocol may be improved by creating error free, reliable, point-to-point links with the data link control (DLC) layer. For this purpose, Bhargavan proposed the MACAW protocol where an acknowledgment (ACK) packet is transmitted after each successful transaction. Including an extra packet in the transaction increases the overhead, which decreases the throughput. However, it is shown in that, for radio channels, the guiding throughput exceeds the increase in overhead. This result may also apply to UWA channels. The MACAW protocol ignores power control and asymmetries that occur. Its performance under power control needs to be investigated. Also, the effect of adding more overhead to the protocol in an environment where propagation delays are excessive needs to he addressed.

Automatic Repeat Request Methods
Automatic Repeat Request (ARQ) is used to detect errors in the data link control layer and then to request the retransmission of erroneous packets. The simplest ARQ scheme that can be directly employed in a half-duplex UWA channel is a stop-and- wait ARQ where the source of the packet waits for an ACK from the destination node for the confirmation of’ error-free packet transmission. Since the channel is not utilized during the round trip propagation time, this ARQ scheme has low throughput. In go-back-N and selective repeat ARQ schemes, nodes transmit packets and receive ACKs at the same time and therefore require full duplex links. Dividing the limited bandwidth of the UWA channels into two channels for full duplex operation can significantly reduce the data rate of the physical layer. However the effect on the overall network throughput needs to be investigated. The selective repeat ARQ scheme can be modified to work on half duplex UWA channels. Instead of acknowledging each packet individually at reception time, the receiver will wait for N packet durations and send an ACK packet with the id numbers of packets received without errors. Accordingly the source of the packets will send N packets and wait for the ACK. Then the source will send another group of N packets that contains the acknowledged packets and new packets.

Acknowledgements can be handled in two possible ways. In the first approach which is called positive acknowledgement, upon reception of an error-free, the destination node will send an ACK packet to source node. If the source does not receive an ACK packet before preset time out duration, it will retransmit the data packet. In the case of a negative acknowledgement, the destination sends a packet if it receives a corrupted packet or does not receive a scheduled data packet. A negative acknowledgement may help to 

conserve energy by eliminating the need to send explicit ACK packets and retransmission of data packets in case of a lost ACK packet. When combined with a MACA type MAC protocol, the negative acknowledgement scheme may provide highly reliable point-to-point links due to the information obtained during RTS-CTS exchange.

A realization of underwater acoustic networking is the U.S. Navy’s experimental Telesonar and Seaweb program. Telesonar links interconnect distributed underwater nodes, potentially integrating them as a unified resource and extending naval net centric operations into the underwater battle space. Seaweb provides a command control, communications, and navigation infrastructure for coordinating autonomous nodes to accomplish given missions in arbitrary ocean environments. More generally Seaweb networking is applicable for oceanographic telemetry, underwater vehicle control, and other uses of underwater wireless digital communications.

Telesonar and Seaweb experimentations address the many aspects of this problem including propagation, signaling, transducers, modem electronics, networking command-centre interfacing and transmission security. The major sea tests have included Seawebs ‘98,’99 and 2000.

Experiment Objectives and Approach
Telesonar acoustic links from the digital network of fixed and mobile nodes. Operational objectives mandate reliability, energy efficiency, deploys ability interoperability flexibility, affordability and security. Thus, telesonar links must be environmentally and situationally adaptive, with provision for bidirectional asymmetry. The Seaweb backbone is a set of autonomous stationary nodes (e.g. sensor nodes, repeater nodes and master nodes) collect data from the sensor nodes and forward to the gateways and vice-versa. Seaweb peripherals include mobile nodes (e.g. AUVs).

Seaweb gateways connect with command centers submerged afloat, ashore and aloft including access to terrestrial, airborne, and space-based networks. For example the telesonobuoy serves as a radio / acoustic interface permitting satellites and maritime aircraft to communicate submerged autonomous systems. Similarly submarines can access off board systems with telesonar signaling. A Seaweb server resides at the manned command centers and is an interfaced underwater network.

Seaweb development involves periodic concentration of resources in prolonged ocean network. The annual Seaweb experiments are designed to validate system analysis purposively evolve critical technology areas such that the state of the art advances with greater reliability, functionality and quality of service. The objective of the Seaweb experiments is to implement and test telesonar modems in networked configurations where various modulation and networking algorithms can be exercised, compared and conclusions drawn. In the long term, the goal is to provide for a self-configuring network of distributed nodes, with the network links adapting to the prevailing environment through automatic selection of the optimum transmit parameters.

Initialization and Routing
Since the network in consideration is an ad hoc network, an initialization algorithm is needed to establish preliminary connections autonomously. This algorithm is based on polling and as such it guarantees connectivity to all the nodes that are acoustically reachable by at least one of their nearest neighbors. During initialization the nodes create neighbor tables. These tables contain a list of each nodes neighbors and a quality measure of their links, which can be received SNR from the corresponding neighbor. The neighbor tables are then collected by the master node and a routing tree is formed. Optimum routes are determined with the help of a genetic algorithm-based routing protocol. The routing protocol tries to maximize the lifetime of battery-powered network by minimizing the total energy consumption of the network. The minimum energy required to establish reliable communication between two nodes is used as the link distance metric. A master node collects the link cost information from the network nodes, determines optimum roots, and sends the routing information hack to the nodes. The optimization algorithm favors multihop links to the expense of increased delay.

The performance of acoustic links between nodes can degrade and even a link can he permanently lost due to a node failure. In such cases the network should be able to adapt itself to the changing conditions without interrupting the packet transfer ibis robustness can be obtained by updating the routes periodically.

In the current design the master node creates a routine tree depending on neighbor labels reported by its nodes. If a node reports that a link performance has degraded or it is no longer available, the master node selects new routes that take the place of the failed link. The changes in the routing tree are reported to all related nodes. This procedure ensures that nodes won’t attempt to use a failure link. In this way unnecessary transmissions that increase battery consumption are avoided.

Media Access Protocol
The media access protocol for Seaweb is based on MACA protocol which uses R CTS-DATA exchange. The network employs the stop-and-wait ARQ scheme. If the source cannot receive CTS from the destination after a predetermined time interval it repeats RTS. If after K trials of RTS, the source cannot receive CTS it decides that link is no longer available and returns to low power state. If the source receives CTS, it immediately transmits the data packet. The RTS / CTS exchange is used to determine the channel conditions, and this information is used to set the acoustic modem parameters such as output power level. An ACK signal is send by the destination upon receipt of a correct data packet to provide positive acknowledgment to the source in the data link layer. The protocol can also handle negative acknowledgments depending on the operation mode selected by the user. Figure 3 illustrates the MAC protocol.

If two nodes send an RTS to each other, unnecessary retries may occur because both nodes will ignore the received RTS command. Each node will then wait for another node to send CTS for time-out duration and retransmit their RTS packet. This problem is solved by assigning priority to the packets that are directed towards the master nodes.

An overview of basic principles and constraints in the design of reliable shallow water acoustic networks that may be used for transmitting data from a variety of undersea sensors to onshore facilities. Major impediments in the design of such networks are considered including severe power limitations imposed by battery power, severe bandwidth limitation, channel characteristics including long propagation times, multi path, and signal fading.

Multiple access methods, network protocols and routing algorithms are also considered. Of the multiple access methods considered it appears that CDMA achieved by either frequency hopping or direct sequence provides the most robust method for the underwater network environment. Currently under development are modems that utilizes these types of spread-spectrum signals to provide multiple access capability to the various nodes in the network. Simultaneous with current modem development there are several investigations on the design of routing algorithm and network protocols.

The design example of the shallow water network employed in Seaweb embodies the power and the bandwidth constraints that are so important in digital communication through underwater acoustic channels. As an information system compatible with low bandwidth, high latency and variable quality of service, Seaweb offers a blueprint for the development of future shallow water acoustic networks. Over the next decade, significant improvements are anticipated in the design and implementation of shallow water acoustic networks as more experience is gained through at-sea experiments and network simulations.

1 comment:

leave your opinion