ES2321152A1 - Distributed routing protocol based on clusters for wireless submarine sensors networks. (Machine-translation by Google Translate, not legally binding) - Google Patents

Abstract

Distributed routing protocol based on clusters for wireless submarine sensor networks. Wireless submarine sensor networks present many challenges due to the properties of underwater environments, such as long propagation delays, node mobility and limited bandwidth. It is essential to design efficient routing protocols with respect to energy because the sensor nodes are powered by batteries that are difficult to load or replace. Most terrestrial sensor network protocols can not be applied because they have been designed for stationary networks and the protocols for existing ad-hoc networks do not work because they use flood techniques or continuous exchange of routing messages. A routing protocol without gps is proposed that minimizes the exchange of routing messages and uses aggregation of data. It assumes mobility of nodes and compensates for propagation delays using temporary advances combined with guard times. (Machine-translation by Google Translate, not legally binding)

Description

Distributed Routing Protocol Based in clusters for wireless underwater sensor networks.

Technical sector

The following invention consists of a protocol of cluster-based distributed routing for networks of wireless underwater sensors (Underwater Wireless Sensor Networks, UWSNs). It falls within the technical sector of design and evaluation of network protocols in the area of engineering telematics

State of the art

Most routing protocols proposed for terrestrial sensor networks cannot be successfully applied in wireless underwater sensor networks because they have been designed for stationary networks and frequently they use techniques based on flooding of packages (flooding).

The routing protocols for multisalt ad-hoc networks are not suitable because they are based on a continuous exchange of signaling messages (proactive ad-hoc routing) or use a route discovery procedure based on the flood technique ( ad-hoc routing reagent); These mechanisms are inefficient in large-scale submarine networks because they consume excessive resources in terms of bandwidth and energy. On the other hand, geography - based ad-hoc routing protocols cannot currently be applied to underwater environments because it is not clear how sensor nodes can obtain accurate information without excessive power consumption; GPS (Global Positioning System) is not helpful to achieve this purpose, as it uses radar waves in the 1.5 GHz band and these waves do not propagate in water
Marine.

To avoid the inconvenience of these mechanisms, the holder of the present invention, Dr. Mari Carmen Domingo Aladrén, has developed a new protocol for distributed routing based on clusters, whose characteristics Fundamental are the following:

The routing protocol developed takes into account energy conservation, it has been designed to long-term aquatic monitoring applications and not reviews temporarily, and you don't need to correct it Operation support GPS. This routing protocol does not use flood techniques, minimize message exchange proactive and uses data aggregation to eliminate information redundant before it is transmitted to the base station (sink)

It is proposed to use TDMA (Time-Division Multiple Access) and CDMA (Code Division Multiple Access) with DSSS (Direct Sequence Spread Spectrum) using pseudo-orthogonal codes for communication within a cluster and only CDMA with DSSS using pseudo-orthogonal codes in all other communication processes. Clusters adjacent to another cluster use different spreading codes and spatial reuse of codes provides scalability. This eliminates interference within a cluster and reduces interference between clusters. In addition, our routing protocol compensates for high propagation delays in the aquatic environment using a continuously adjusted time advance combined with guard times to minimize data loss and maintain communication quality.
tion

Detailed description of the invention A. Protocol architecture

In this section a new protocol is introduced of adaptive routing and able to organize itself that form clusters using a distributed algorithm.

To develop this protocol, they have made a series of assumptions about the sensor nodes submarines and the underlying network model:

1) The underwater sensor nodes move randomly

2) All underwater sensor nodes are energy restricted and have different battery capacities maximum.

3) The underwater sensor nodes always They have data ready to be sent to the base station. All nodes have similar calculation capabilities and processing, and are capable of performing processing functions signal

4) All underwater sensor nodes they can perform power control to adjust their power transmission.

The first assumption motivates the need for design a new routing protocol that is robust against to the breaking of a link due to mobility. The second and third assumptions motivate the need to develop a routing protocol that takes conservation into account of energy

The operation of this protocol of routing is illustrated in Fig. 1. The nodes organize themselves same in local clusters and a node is selected as Unique cluster-head for each cluster. All the no cluster-heads nodes transmit their data to their cluster-head in a single jump while the node cluster-head receives data from all members of the cluster, performs signal processing functions on the data (for example, data aggregation) and transmits this data to the source (through other cluster-heads that act as intermediate nodes) using multi-hop routing. Nodes next to each other process data very frequently correlated because they monitor the same phenomenon and with the help of aggregation techniques effective data not redundant can be extracted by the cluster-head and sent to the base station. This saves energy. The cluster-heads are responsible for coordination between nodes within clusters and communication with each other.

This routing protocol incorporates random rotation of the cluster-head between sensors to avoid draining the battery of any underwater sensor that belongs to the network. In this way, both the Energy consumption as traffic load.

The operation of this routing protocol It is divided into rounds (see Fig. 2). All clusters are formed during the establishment phase or creation process of clusters and data transfer takes place during the phase stable or network operation phase.

During the network operation phase several frames are sent to the cluster-head; a plot is formed by several data messages that nodes do not cluster-heads send to cluster-head using a schedule (each non-cluster-head sensor sends a message from data consuming a slot time). Later, both phases are Repeated periodically. You must ensure that the operation phase network is long compared to the process of creating clusters to minimize signaling and improve the performance of the routing protocol

B. Algorithm of selection of cluster-heads

The selection algorithm of cluster-heads allows you to choose a certain number of cluster-heads during the process of creating clusters

The cluster-head selection is based on two parameters:

A main parameter (residual energy of node) is used to select an initial group of cluster-heads, and a secondary parameter (cost) is Use for conflicting cases. A conflicting case occurs when two nodes within the same range with respect to each other, announce his willingness to become cluster-heads.

A node initially establishes its probability of becoming cluster-head of the following shape:

one

where C i represents the current battery level (residual energy of the node), and C MAX represents the maximum battery capacity. C prob is a constant small fraction used to establish an initial percentage of cluster-heads, thus limiting the number of nodes that proclaims itself as cluster-heads. CH prob is not allowed to be less than a small probability called p min. This restriction is necessary to increase the probability that some nodes are proclaimed cluster-heads even if the battery levels of all sensors are low. to ensure that the routing protocol will continue to function correctly.

The second parameter, called cost, is a estimate of the cost of communication, which is a function of the Distance between source and destination. Each node does not cluster-head decides which cluster you want to belong to choosing that cluster-head that requires the minimum communication energy to reach it, and in consequently, the level of power required for transmission (we assume that a node can control power). The transmission power is directly proportional at the distance between sensor nodes in scenarios with little water deep (with a marine depth of less than 100 m). For the therefore, each non-cluster-head node must calculate its distance (cost) to each neighboring cluster-head with Method assistance based on acoustic arrival time (acoustic-only Time-of-Arrival (ToA)) (by example, measuring round trip time (round-trip time) experiencing a signal acoustics) and select the cluster-head more next.

During the process of creating clusters, the nodes examine their excess energy level and calculate their probability of becoming a cluster-head CH prob. If CH prob is above a random value between 0 and 0.5, that node chooses itself as a cluster-head and sends an announcement message to its neighbors using CDMA.

During this phase, each sensor node processes the Ads received by cluster-heads and select the cluster-head with lower cost. After that each node has chosen which cluster it wants to belong to, You must inform the cluster-head. Each node transmits a membership request message to that cluster back towards the cluster-head chosen using CDMA. This message It is short and contains the node identifier and the identifier of the cluster-head. If a node has not received any announcement by some cluster-head, must Send your data directly to the base station.

As a result, the nodes with the highest residual energy are those that are proclaimed as cluster-heads and a lower cost is spent on communication. Furthermore, the calculation of CH {prob} allows heterogeneous nodes with different maximum capacities of battery involved in the network equal.

The expected number of nodes cluster-head using this algorithm is:

2

C. Cluster formation algorithm and operation phase network

After the cluster creation process, each cluster-head knows which nodes belong to its cluster. Now the cluster-head must coordinate the data transmissions in its own cluster. He cluster-head sets a schedule with TDMA (Time-Division Multiple Access) and transmit this planning using CDMA to the members of the cluster. TDMA has been selected as MAC protocol (Medium Access Control) within a cluster because it avoids collisions between members that are not cluster-head and because it allows that no cluster-head nodes are inactive while not transmitting; therefore they remain in mode asleep (sleep) and thus reduce energy consumption. In the Fig. 3 we can see an example of the coordination of the transmission of data in a cluster with 3 member nodes (in addition to the cluster-head). Cluster head plan to which node it belongs to transmit each time and when It will begin its transmission. Then send the planning to the cluster members.

A major problem in communications underwater is the fact that data messages from different cluster members could overlap in the cluster-head due to the tall and different propagation delays that they experience in the underwater environment, resulting in low transmission quality and even losses of communications. The proposed solution is that each sensor node advance your transmission in relation to your reception a time to compensate the propagation delay. This value is called advance temporal advance, a concept used in other systems of communications such as GSM (Global System for Mobile Communications). The timing advance value for each node can be calculated only by the cluster-head, and then it sends it to the underwater sensor nodes of the network included in TDMA planning.

When a cluster-head knows what nodes will belong to your cluster, send them an acoustic signal with in order to measure the round trip time (round-trip) and as a result estimate the delay of propagation for each no cluster-head node in its cluster with the help of ToA (Time of Arrival) techniques.

Assume that the nodes N 1, N 2, N 3 ... N f have been associated with the cluster with cluster-head CH j; then the cluster-head sends acoustic signals to know the propagation delays of itself to each member of the cluster, which are \ tau_ {1}, \ tau_ {2}, \ tau_ {3} ... \ tau_ {f } respectively.

The cluster-head node checks Which node has the maximum propagation delay:

3

where NCH is the subset formed by the non-cluster-head nodes that belong to the cluster with cluster-head CH j.

The cluster-head knows that once the schedule has been sent to the cluster members, the node N l with the greatest propagation delay will receive it only after \ tau_ {l}; therefore it establishes (once the schedules have been sent) a reference or start time for transmission after \ tau_ {l}, (reference starting moment); this means that the nodes N 1, N 2, N 3 ... N f, which receive their schedules only after ta 1, ta 2 , \ tau_ {3} ... \ tau_ {f}, they must wait once they have received their schedules for a time \ tau_ {1} - \ tau_ {1}, \ tau_ {1} - \ tau_ {2}, \ tau_ {1} - \ tau_ {3} ... \ tau_ {l} - \ tau_ {f}; Thus, we ensure that all nodes adjust the same start time as a reference at the same time. The nodes must process the schedule using t proc, although processing times are negligible compared to propagation delays. In the example of Fig. 3 the cluster-head has sent acoustic signals to estimate the propagation delay of each node towards itself, which are \ tau_ {1}, \ tau_ {2} and \ tau_ {3} for nodes N 1, N 2 and N 3, respectively. The cluster-head sets a start time for transmission or suitable reference for all nodes, considering that the N {3} node has the propagation delay greater and therefore receive planning after \ tau {3} .

The frames are sent to cluster-head; the frames are divided into slots of time and each time slot is used to transmit a message from data from a different node;

The duration of a frame is set to k \ times T slot, where k represents the number of time slots in a frame and this value is equal to the number of members of a cluster that are not cluster-head (the number of members of a cluster minus the cluster-head); the value k varies depending on the number of members of a cluster that are not cluster-head in each cluster; if on average there are \ frac {\ mathit {N}} {\ mathit {z}} nodes per cluster, k = \ left (\ frac {\ mathit {N}} {\ mathit {z}} - 1 \ right) .

A round can be defined as a period of time where clusters are organized and frames are transmitted from the cluster members to the cluster-head. A frame consists of a series of data messages that each non-cluster-head node sends to the cluster-head using a schedule (each non-cluster-head sensor node sends a data message consuming a temporary slot). When the first frame is sent, the number of shifts t is initialized to 0 and each node must send a data message on this first turn. The nodes are classified according to their propagation delays from the smallest to the largest and it has been decided that the nodes should send information in this order, because in this way the cluster-head should only wait for \ tau_ {S} seconds (minor propagation delay) the transmission of the first frame and henceforth the frames are sent uninterruptedly. This means that in this planning node with minimum propagation delay \ {S} tau should start sending a data message at T {inicio_ {s}} (initial time reference) consuming a time slot. The node with the second minimum propagation delay is as follows and so the process continues. With this transmission order a certain extra delay is saved.

The second node N s + 1 must begin the transmission of its first data message at:

4

where T slot_ {s} means transmission time (using a temporary slot) of the first data message of node N_ {s}.

The third node N s + 2 must begin its transmission at:

5

In general, node N s + i, must begin transmitting its first data message at:

6

Finally, the last node N s + u transmits its first data message. When all nodes have sent a data message, a full frame is transmitted; the number of turns t is increased by 1, i is initialized to 0 for each new turn and the process is repeated in the same order of transmission. In general, node N s + i, must begin the transmission of its ( t + 1) -th message of data in order in:

7

where u represents the number of nodes that transmits minus 1.

We assume that the transmission times of the data messages are long enough so that each instant of transmission T start_ {s + 1} \ geq 0.

In the example of Fig. 3 nodes are classified according to their propagation delays, from the node with the propagation delay less to the node with the propagation delay greater in the order N {1}, N {2} and {3} N, and should send frames in this order; thus decreases the reception time of the first frame.

So far it has been assumed that the delay of propagation is the same for each data message sent by a sensor node in particular, but in reality the delay of propagation varies due to channel fluctuations caused by the relative movement of the transmitter, receiver or the surfaces of scattering Therefore, in order to avoid collisions between two members of a cluster whose data messages arrive simultaneously to the same cluster-head, we have decided that each node reserves a period of time called "guard time" of the duration of its transmission. This period allows the information to reach a certain distance without that there is interference caused by some subsequent transmission. The establishment of a temporary advance reduces considerably the length of guard time, which has a duration of one tenth of the temporary slot.

After having transmitted a plot to cluster-head, it is required to consume a time of slot to perform cluster maintenance; the nodes go changing position over time due to mobility and it is necessary to modify the members of a cluster as well as the TDMA schedules in accordance with these changes. He Cluster maintenance algorithm operation is explained in the next section.

When a cluster-head has received a data message from each member of its cluster, it performs signal processing functions to compress that data together with its own data message into a single signal. The composite signal is sent to the base station through other cluster-heads using CDMA and multi-hop routing. Each cluster-head CH i selects as next hop to forward its frames to the base station S the adjacent cluster-head CH j (which minimizes the distance CH i - CH j) and satisfies:

distance CH i - CH j <distance CH i - S \ hskip0.5cm and \ hskip0.5cm distance CH i - S <distance CH i - S

CH i knows the distance CH j- S because the announcement message sent by CH j during the establishment phase includes this information. The distances are calculated with the help of ToA techniques.

       \ vskip1.000000 \ baselineskip
    

D. Cluster Maintenance

The sensor nodes in acoustic environments submarines move and consequently the organization of clusters is altered as well as TDMA planning. The algorithm of Cluster maintenance would work as follows:

During the temporary slot reserved for maintenance purposes every node does not cluster-head should estimate its propagation delay to your cluster-head in your cluster with the help of ToA techniques. If this delay is β% greater than the same parameter calculated during the establishment phase, the node must re-estimate the distance to each neighbor again cluster-head that had announced itself during the establishment phase using ToA techniques and then you must select the neighboring cluster-head that is closer. If a cluster-head is now chosen different from the previously selected cluster-head, the node must send a message to join the new cluster and another message to leave the previous cluster. Then the affected cluster-heads should recalculate the TDMA schedules and send them to the members of your clusters

       \ vskip1.000000 \ baselineskip
    

E. MAC protocols

Limited bandwidth and high delays propagation of the underwater acoustic channel impose restrictions severe for the design of MAC protocols (Medium Access Control). It is proposed to use TDMA and CDMA (Code Division Multiple Access) with DSSS (Direct Sequence Spread Spectrum) using codes pseudo-orthogonal for communication within a cluster and only CDMA with DSSS using codes pseudo-orthogonal in all other processes of communication.

It has been assumed in the design of the protocol of routing that each node has a correlator filter bank to obtain data from different spreading codes (spreading) A multi-user receiver bank is capable of withstand high interference levels to obtain data from Different spreading codes.

The operation of the protocol routing can be described as follows:

Each node that chooses itself as cluster-head, randomly select a single spreading code to send warning messages to your neighbors. Yes a cluster-head candidate just heard some previous announcement message from another cluster-head, use another unique and different spreading code. If a node receives two warning messages from two cluster-heads different that use the same spreading code, the node must notify one of them about the situation so that it resends again an announcement message using a new spreading code different than invalidates the previous one. If there are no more codes spreading available, it is not possible for a node to declare itself same as cluster-head.

One of the advantages of using a protocol based in clusters is that a different spreading code is needed by cluster and not per node; therefore, the number of codes is not It extinguishes so quickly. Clusters adjacent to another cluster use different spreading codes and scalability is achieved by spatial reuse of them.

When a node replies to an announcement message of a particular cluster-head to join your cluster, use the same spreading code that you used at the time the cluster-head All nodes not cluster-heads send their data to cluster-head using TDMA and the same code of spreading, and again the same code is used when the cluster-head sends its aggregated data to the base station using multi-hop routing.

The advantages of using CDMA / TDMA are that the interference within a cluster is eliminated and interference between clusters is reduced with the allocation of a code based in the transmission. Interference within a cluster is eliminated because no cluster-heads nodes transmit in order using a TDMA scheme. Interference between clusters is reduce because clusters adjacent to another cluster use a code of different transmission.

Brief description of the figures

To improve the compression of how much is left described herein, a series of figures are attached. Figures Fig. 1 and Fig. 2, have already been discussed previously in the description of the new routing protocol.

In the figure Fig. 3 previously commented we can see an example of the clustering algorithm and of the network operation phase with a cluster-head and three cluster member nodes. Figures Fig. 4 and Fig. 5 they illustrate an example of code assignment between clusters. The Fig. 5 shows the structure of a frame for the network partition illustrated in Fig. 4.

Exhibition of an embodiment

The operation of this protocol routing is illustrated in figures Fig. 3, Fig. 4 and Fig. 5. In Fig. 4 it can be observed how during the phase of establishment of the routing protocol the submarine network has It is divided into 8 clusters, each of which consists of a cluster-head and three or four members not cluster-heads. During the network operation phase in each cluster each of its members transmits messages from data to the cluster-head using the same spreading code than the other members. The frames that a cluster-head are made up of data messages, specifically each data message belongs to a member different from the cluster. As an example we can see that in Fig. 3 the members of a cluster transmit information to the cluster-head following a specific order (from the node with the lowest propagation delay to the node with the highest propagation delay). The start of each shipment is planned data message to prevent data messages from colliding in the cluster-head and to get the cluster-head continuously receive data messages and in this way the bandwidth is better used.

Fig. 5 represents the structure they would have some of these data messages that are transmitted to the cluster-head (in addition to the data a guard time) and the structure that would have a typical plot that would be sent by the members of a cluster to a cluster-head Finally the cluster-head adds its own data message to the frame received, performs data aggregation to eliminate the redundancy and sends the resulting frame to the base station to through a multi-route route that crosses several cluster-heads. After the cluster-head has received a plot from members of your cluster, each node must dedicate a slot of time to cluster maintenance (with the operations that this entails).

Claims (4)

1. Distributed routing protocol, based on clusters, for wireless underwater sensor networks, which organizes the network into clusters and selects a cluster-head for each cluster characterized by:

sending information by each node in a single jump to the cluster-head and for the transmission of data from that cluster-head, using multisalt routing between cluster-heads to the base station (sink),

the selection of node that acts as a cluster-head following a random rotation between sensors,

the choice of membership in a cluster by each node not cluster-head, choosing that one cluster-head that requires the minimum energy of communication to reach the node.

2. Routing protocol according to claim 1, characterized by using data aggregation to eliminate redundant information before it is transmitted to the base station (sink). 3. Routing protocol according to previous claims characterized by using TDMA (Time-Division Multiple Access) and CDMA (Code Division Multiple Access) with DSSS (Direct Sequence Spread Spectrum) using pseudo-orthogonal codes for communication within a cluster and only CDMA with DSSS using pseudo-orthogonal codes in all other communication processes. Clusters adjacent to another cluster use different spreading codes and spatial reuse of codes provides scalability. 4. Routing protocol according to previous claims characterized by compensating the high propagation delays of the aquatic environment using a continuously adjusted time advance combined with guard times.