Module 4 QB Solutions

1. Elaborate on the concepts of Energy Efficient Unicast Routing Protocol.

Ans. Minimize energy per packet (or per bit): The most straightforward formulation is to look at the total
energy required to transport a packet over a multihop path from source to destination (including all
overheads). The goal is then to minimize, for each packet, this total amount of energy by selecting a good
route.

Maximize network lifetime: A WSN’s task is not to transport data, but to observe (and possibly control).
Hence, energy-efficient transmission is at best a means to an end and the actual end should be the
optimization goal: the network should be able to fulfill its duty for as long as possible.
Which event to use to demarcate the end of a network’s lifetime is, however, not clear either. Several options
exist
• Time until the first node fails.
• Time until there is a spot that is not covered by the network (loss of coverage, a useful metric only for
redundantly deployed networks).
• Time until network partition (when there are two nodes that can no longer communicate with each other).

Routing considering available battery energy: While maximizing the network lifetime is clearly a useful
goal, it is not immediately obvious how to reach this goal using observable parameters of an actual network.
As the finite energy supply in nodes’ batteries is the limiting factor to network lifetime, it stands to reason
to use information about battery status in routing decisions. Some of the possibilities are:
• Maximum Total Available Battery Capacity: Choose that route where the sum of the available battery
capacity is maximized, without taking needless detours.
• Minimum Battery Cost Routing (MBCR): Instead of looking directly at the sum of available battery
capacities along a given path, MBCR instead looks at the “reluctance” of a node to route traffic. This
reluctance increases as its battery is drained.
• Min–Max Battery Cost Routing (MMBCR): This scheme follows a similar intention, to protect nodes
with low energy battery resources. Instead of using the sum of reciprocal battery levels, simply the
largest reciprocal level of all nodes along a path is used as the cost for this path. Then, again the path
with the smallest cost is used. In this sense, the optimal path is chosen by minimizing over a maximum.
• Conditional Max–Min Battery Capacity Routing (CMMBCR): Another option is to conditionalize
upon the actual battery power levels available. If there are routes along which all nodes have a battery
level exceeding a given threshold, then select the route that requires the lowest energy per bit. If there
is no such route, then pick that route which maximizes the minimum battery level.
• Minimize variance in power levels: To ensure a long network lifetime, one strategy is to use up all the
batteries uniformly to avoid some nodes prematurely running out of energy and disrupting the network.
Hence, routes should be chosen such that the variance in battery levels between different routes is
reduced.

Minimum Total Transmission Power Routing (MTPR): Without actually considering routing as such,
Bambos looked at the situation of several nodes transmitting directly to their destination, mutually causing
interference with each other. A given transmission is successful if its SINR exceeds a given threshold. The
goal is to find an assignment of transmission power values for each transmitter (given the channel attenuation
metric) such that all transmissions are successful and that the sum of all power values is minimized. MTPR
is of course also applicable to multihop networks.
2. Explain Rumor Routing with an example and a neat diagram.
Ans. Assume some sensors are interested in certain events (e.g. temperature exceeding a given value) and a
sensor can observe it. Classical options are to flood either the query for the event or the notifications that an
event has occurred through the entire network. The “rumor routing” approach proposed here does not flood
the network with information about an event occurrence but only installs a few paths in the network by
sending out one or several agents. Each of these agents propagates from node to node and installs routing
information about the event in each node that it visited. This is illustrated in Figure 11.2(a) where the node
in the middle detects an event and installs two event paths in the network (shaded areas). Once a node tries
to query an event (or to detect whether an event actually exists), it also sends out one or more agents. Such
a search agent is forwarded through the network until it intersects with a preinstalled event path and then
knows how to find an event.

In Figure 11.2(b), the node in the lower left corner sends out such a search, which happens to propagate
upward until it intersects with one event path. All these agent propagations are limited to avoid endless
circling of data. The rationale behind this technique is the relatively high probability that two random lines
in a square intersect each other. While neither the event paths nor the search paths will in reality be straight
lines, the approximation is claimed to be good enough. Using five instead of one event paths increases this
probability to about 99.7 %. In effect, rumor routing allows to trade off effort in path creation and/or search
against probability of detecting an event. There are a few more functionalities included in rumor routing.
For example, agents spread information about more than one event if they have crossed an event path for
another event. Also, an agent uses opportunities to shorten existing event paths if they know about shorter
paths.

3. Explain the Multipath unicast routing.

Ans. Multipath routing protocols construct several paths between a given sender and receiver.
Sequential Assignment Routing (SAR): As a basic rule of thumb, computing such k-disjoint paths requires
about k times more overhead than a single-path routing protocol. The multipath-induced overhead was
reduced by focusing the disjointness requirements to that part of a network where they truly matter – near
the data sink, as the nodes close to the sink are (often) those that likely are going to fail first because of
depleted battery resources. Hence, they only require paths to use different neighbors of the sink. The
Sequential Assignment Routing (SAR) algorithm achieves this objective by constructing trees outward from
each sink neighbor; in the end, most nodes will then be part of several such trees. A packet’s actual path is
then selected by the source on the basis of information about the available battery resources along the path
and the performance metrics (e.g. delay) of a given path.
Constructing energy-efficient secondary paths: When using multiple paths as standby paths to quickly
switch to when the primary path fails, an obvious concern is that of the energy efficiency of these secondary
paths compared to the optimal primary path. The first observation is that strictly requiring node disjointness
between the various paths tends to produce rather inefficient secondary paths as large detours can be
necessary. To overcome this problem and yet retain the robustness advantages of multiple paths, they suggest
the construction of so-called “braided” paths (sometimes also called “meshed” multipaths). These braided
paths are only required to leave out some (even only one) node(s) of the primary path but are free to use
other nodes on the primary path. This relaxed disjointness requirement results in paths that can “stay close”
to the primary path and are therefore likely to have a similar, close to optimal energy efficiency as the
primary path. Figure 11.5 illustrates these two redundant paths’ concepts.

Simultaneous transmissions over multiple paths: When using multiple paths as a standby for the primary
path, failover times might be improved compared to strictly single-path solutions. Nevertheless, there is
some delay in detecting the need to use a secondary path. Depending on which node makes this decision –
only the source node or any node on the primary path – there can be more or less overhead involved. To
further shorten the time to delivery and to increase the delivery ratio of a given packet, it is also conceivable
to use all or several of the multiple paths simultaneously. The simplest idea is to assume node-disjoint paths
and to send several copies of a given packet over these different paths to the destination. Clearly, this trades
off resource consumption against packet error rates. The degree of redundancy and the number of paths can
be tuned to the expected error behavior, trading off overhead against residual packet error rate.

Randomly choosing one of several paths: When maintaining multiple paths, it actually makes sense also
to use paths that are less energy efficient than the optimal one. One reason to do so is to share the load among
all nodes in order to use the available battery capacity in the network better. Each node maintains an energy
cost estimate for each of its neighbors (toward the destination, packets are not routed “away” from their
destination). When forwarding a packet, the next hop is randomly chosen proportional to the energy
consumption of the path over this neighbor. To the upstream node, the appropriately weighted average of
these costs (i.e., the harmonic mean of the costs) is reported.

Trade-off analysis: Clearly, supporting such multiple paths in a network implies a trade-off between
robustness (the probability that paths are available even after node failure) and energy efficiency (as both
the management of these paths and the nonoptimal choices made for packet forwarding decisions imply
increased energy expenditure) – irrespective of the concrete routing protocol in use.
4. Explain the Overview of possible multicast approaches (sep).
Ans.

Source-based tree: The first idea is to construct, for each source, a tree, rooted at the given source, that
contains all the destinations for this source and, if necessary, additional nodes of V to ensure that the tree
can be constructed.
Which tree to select (out of, in general, many possible ones) is determined by the optimization
goal, which reflects the link costs:
For each source, minimize the total cost: Try to find a tree for which the sum of all link costs is minimal
(over all possible trees rooted at the source). This is the Steiner tree problem and matches best the intuitive
expectation how a multicast routing structure should look like. It is, however, NP-complete.
For each source, minimize the maximum cost to each destination: Owing to the complexity
of the previous optimization problem, a different optimization goal can be considered: Instead of trying to
minimize the total cost of the tree, one can minimize the costs to each individual destination separately. In
effect, this maps the multicast problem to repeated unicast shortest path problems, which can be solved by
any routing algorithm, for example, Dijkstra’s.

Shared, core-based tree: Constructing and maintaining a dedicated tree for each source incurs considerable
overhead. This overhead can be reduced if only a single tree is maintained – typically, this is promising when
the destination sets for all sources are identical. The obvious downside is that for a given source the paths to
its destinations can, in general, no longer be as short as with a dedicated, source-based tree. To share a tree
among several sources, a representative node in the network (not necessarily a source or a destination node)
is selected and, from this node, a tree is constructed to contain all destination nodes; this tree is shared among
all the sources. Selecting this core node is again NP-complete and optimization goals similar to the source-
based trees can be considered. In such a shared tree concept, the core node evidently becomes a single point
of failure. To overcome this shortcoming, multicore shared trees are also considered in the literature.

Mesh: While trees represent the overhead-optimal routing structures, they are not redundant – failure of
even a single link will disconnect the tree. Adding additional links to the tree to obtain redundancy, however,
will alter its essential properties, in particular the absence of cycles. The resulting routing structure will be a
mesh and requires more complicated forwarding structures than does a simple tree.
5. Explain the BIP algorithm that exploits the wireless multicast advantage.
Ans. A node that wants to transmit to multiple neighbors (because it has multiple children in the tree, for
example) experiences a proportional cost. The Broadcast Incremental Power (BIP) algorithm [874] differs
here in that it exploits the wireless multicast advantage to compute a heuristic for a broadcast tree. The core
idea is that a node that is already transmitting to some other node would only have to raise its transmission
power in order provide data also to further nodes, without incurring cost for another transmission. Hence,
the additional cost for a node to supply a further node with data is only the difference between the current
and the needed (higher) transmission power.
The broadcast incremental power algorithm for exploiting the wireless multicast advantage

6. Data-centric vs. data aggregation networking.

Ans.

7. Geographic routing: GPSR ,GOAFR,GERAF,GEM.

Ans. Greedy Perimeter Stateless Routing (GPSR) protocol: GPSR forwards a packet as long as possible
using greedy forwarding with the “most forward” rule. If a packet cannot make any more progress, the packet
is switched to another routing mode: perimeter routing. A perimeter is a set of nodes defining a face (the
largest possible region of the plane that is not cut by any edge of the graph; faces can be exterior or interior).
Perimeter routing essentially consists of sending the packet around the face using the right-hand rule. To do
so, the packet carries information where it entered a given face. This node v and the connecting line between
v and the destination are used to decide whether the packet should leave the face and proceed to the next one
(when the edge from the current node to the next node on the face intersects the connecting line between v
and the destination node). Also, the packet can return to greedy forwarding if the distance of the current node
to the destination and node v has been effectively reduced.
Figure 11.13 illustrates how a packet would be routed from node A to node Z. While at node A, the packet
can be greedily forwarded to node D. At node D, greedy forwarding fails (both B and C are further away
from Z than D itself), so the packet has to be routed around the perimeter of the interior face defined by
BFGCD. That is, it is forwarded to B and from there to F. Here, edge FG intersects line DZ and routing can
proceed to the next face (note that greedy forwarding to G would not help here). The packet proceeds around
the perimeter of the exterior face via E and I to H, from there via K to J and thence to L and Z (the last steps
via greedy forwarding).

Greedy and (Other Adaptive) Face Routing (GOAFR)+ algorithm: GOAFR is the worst-case optimal
and at the same time efficient in the average case. The crucial point is when to fall back to greedy mode –
too soon loses worst-case optimality, too late wastes average-case performance. Two techniques realize this
behavior:
• The algorithm maintains a bounding circle, cantered at the destination node, that prevents the face search
from needlessly exploring in the wrong direction. This circle is reduced at every step in the greedy
forwarding phase and can be enlarged in face routing if, with the current circle restrictions, no progress
toward the destination can be made.
• A packet maintains two counters, p and q. When switching to face-based forwarding, both counters are
set to 0. Counter p contains the number of nodes on the face perimeter that are closer to the destination
than is the node where face search started; q counts nodes farther away. The algorithm falls back to
greedy search if p > σq (for some properly chosen constant σ), that is, when substantially more nodes
are closer to the destination on this face than are further away.

Geographic Random Forwarding (GeRaF): Two people investigate the combination of position-
informed, random forwarding and nodes that switch on and off to save energy. Their scenario for
investigation is the following:
Assume nodes are uniformly distributed over the plane, each node knows its position, and each node turns
on or off at arbitrary times; nodes also know their position and that of their neighbors. The goal is to transmit
a message, usually over multiple hops, to a destination node the position of which is also known; the
challenge is the constantly changing topology. Their basic idea is then to use receiver-initiated forwarding:
A node S forwarding a message simply broadcasts it without specifying – in the packet or otherwise – which
node shall forward it. Ideally, the node closest to the destination node T and in range of S will pick up the
message and forward it onward. However, since the neighbors of S do not know which of the other nodes is
currently asleep, a deterministic rule to pick the forwarder would be difficult. Hence, the problem is solved
by a position-informed randomization. Define N annuli with inner radius D − 1 + (i − 1)/N, i = 1, . . .,N and
width 1/N (where D = |ST | is the distance between nodes S and T and the radio range is normalized to 1);
let Ai be the intersection of these annuli with the radio range of S.
Graph EMbedding (GEM): GEM methods utilize the methodology from graph theory. GEM embeds a
graph into a WSN regardless of the physical distribution of the network. The algorithm for GEM uses a
spanning tree to construct the graph, then arranges sensor nodes using coordinates in Virtual Polar
Coordinate Space (VPCS)

8. Multicast routing protocols-Shared and mesh protocols.

Ans. Shared, core-based tree protocols
The challenge in core-based tree multicast protocols lies in finding a good core node. Once this node is
determined, essentially the problem can be reformulated as a source-based tree protocol with the core
node as the source.
To give an idea how such a core search could work, the “merge point formation” is explained:
• Assume there are a few sinks in a network to which data shall be distributed via a core-based
multicast tree. A “merge point” for this tree is to be found.
• To do so, each sink broadcasts advertisement messages indicating its presence; each node in the
network collects these advertisements along with sink identifier and number of hops that the
advertisement took.
• After a certain time, each node that has received more than one sink advertisement broadcasts merge
advertisement messages. These messages are only forwarded by nodes that have heard from fewer
sinks or whose cumulative distance to all sinks is larger.
• Eventually, only one node will not have heard other merge advertisements overruling its own and
will declare itself the merge point.
• This leader election result is then spread through the network.
Core-Based Trees
• When a receiver joins a multicast group, its local CBT router looks up the multicast address and
obtains the address of the Core router for the group.
• It then sends a Join message for the group towards the Core.
• At each router on the way to the core, forwarding state is instantiated for the group, and an
acknowledgment is sent back to the previous router.

Mesh-based protocols
To overcome scalability and robustness issues of tree-based protocols, a structure with higher connectivity
is necessary that can connect multiple sources to their destinations.
The first proposal in this sense was the Core-Assisted Mesh Protocol (CAMP). The mesh, a subgraph of the
original graph, has to contain all sources and destinations and provide at least one path from each source to
each destination. The redundancy of the mesh can actually enable shorter paths in the mesh than would be
possible in a core-based tree; it is, however, up to the forwarding procedure to actually be able to exploit
these shortcuts without resorting to flooding the entire mesh with data.
The energy consumption of mesh-based protocols is, on the average, (f + 1)/2 times larger than that of tree-
based multicast protocols (where f ≤ 2 is the node connectivity, analyzed for nodes laid out in a grid pattern).
Hence, their use in sensor networks requires careful deliberation.
As an example of a (in a sense) mesh-based protocol, let us consider the Two-Tier Data Dissemination
(TTDD).
Here, source nodes detect certain events that have to be forwarded to several mobile sinks.
To do so,
• Each source node, after detecting an event, starts to build a regularly spaced mesh of “dissemination
points”, resulting in rectangular cells of known size l.
• Sinks, flood their queries in their local vicinity with a bounded radius to live. This radius is selected
large enough so that at least one dissemination point must be found within it.
• In this sense, two tiers of forwarding functionality are deployed – global multicasting on a mesh and
local query broadcasting.
• Once a dissemination node receives a query, it forwards it upstream to the source of the data via the
multicast mesh.
• Dissemination nodes keep on forwarding (aggregated) queries toward the source and store information
from where the queries arrived.
• This information is used to send the actual data back to the sinks, implicitly setting up a multicast tree
on top of the multicast mesh.

9. Multicast routing protocols-source based tree protocols.

Ans. Source-based tree protocols: Source-based trees can be constructed in a number of different ways.
• A simple heuristic is - A greedy heuristic – Shortest Path Tree.
• Broadcasting using minimum cost spanning tree – Prim’s algorithm.
• An essential algorithm for broadcast trees and the multicast problem is solved by various Steiner
tree approximations.
These approaches work on a basic graph interpretation of the network.
Solutions that exploit the wireless multicast advantage – in particular, BIP is used.

A greedy heuristic – Shortest Path Tree

• A trivial heuristic for broadcasting and/or multicasting is to compute, to each destination, the shortest
path and overlay all these paths onto a tree as Shortest Path Tree.

• Consider a network with the source node at the center, m nodes p1. . . pm distributed on the unit circle,
and another m nodes q1, . . . , qm each placed on the line connecting the origin with node pi at distance
ε from the origin. (as shown in Figure)
• SPT will send one broadcast packet (using the wireless multicast advantage) to the inner ring at cost ε2
(assuming quadratic path loss), and each node qi forwards at cost (1 − ε)2. The total cost is thus ε2 + m
(1 −ε) 2.
• Having the source node send at power 1 will, on the other hand, distributed the information to all nodes.
• Thus, the approximation ratio is (ε2 + m (1 − ε)2)/1, which converges to m as ε → 0.
• Greedy heuristic does not have a good approximation ratio.

Broadcasting using minimum cost spanning tree – Prim’s algorithm.

• A simple broadcasting algorithm can be based on a minimum cost spanning tree.
• It starts with a tree consisting of the source node, and in |V| − 1 steps adds one node per step to the tree.
The node that is added is the one that has the lowest-cost link to any node already in the tree.
• More formally, given a graph G = (V ,E) with edge weights W : E →R (for example, energy cost for the
edge), the algorithm maintains the tree (VT ,ET ) itself and a set of candidate nodes VC along with one
candidate edge per candidate node.

Steiner tree
Definition: A minimum-weight tree connecting a designated set of vertices, called terminals, in an
undirected, weighted graph.
The tree may include non-terminals, which are called Steiner vertices
or
Given an undirected graph with non-negative edge weights and a subset of vertices (terminals), the Steiner
Tree in graph is an MST “T” of minimum weight that contains all given terminals (but may include additional
vertices).

10. Geocasting : TORA,GAMAER,GEO- GRID,TBF.

Ans. TORA (Temporally Ordered Routing Algorithm) is one of a family of link reversal algorithms for
routing in ad hoc networks. For each possible destination in the ad hoc network, TORA maintains a
destination-oriented directed acyclic graph (DAG). In this graph structure, starting from any node, if links
are followed in their logical direction, the path leads to the intended destination. TORA uses the notion of
heights to determine the direction of each link. Despite dynamic link failures, TORA attempts to maintain
the destination oriented DAG such that each node can reach the destination.
GeoTORA is based on TORA. Geocast delivery using GeoTORA consists of two phases: anycasting phase
[When a node wishes to send a packet to the geocast group, it forwards the packet on any of its outgoing
links, the packet eventually reaches one member of the geocast group] and local flooding phase.

Geocast Adaptive Mesh Environment for Routing (GAMER) provides geocast communication in an ad
hoc network. GAMER adapts to the current network environment by dynamically changing the density of
the mesh. Thus, when nodes are highly mobile, a dense mesh is created; when nodes are moving slowly, a
sparse mesh is created. GAMER improve the transmission accuracy significantly, without increasing the
load on the network. GAMER provides a mesh of paths between the source and the geocast region. When a
link breaks due to the changing topology in an ad hoc network, the redundant paths that exist in the mesh
can be used.

GeoGRID
In this protocol the geographic area of the MANET is partitioned into 2D logical grid. Each cell in a grid is
a square of size d x d. The cells are numbered (x, y) following the conventional x-y coordinate. Each host
still has a unique ID (such as IPaddress). In each cell one host will be elected as the gateway of the cell. The
responsibility of gateway hosts is to propagate geocast packets to neighboring cells. For maintaining the
quality of routes, the gateway host of a grid should be the one nearest to the physical center of the grid. Two
versions of GeoGRID will be mentioned one called flooding-based and the other called ticket-based. In the
flooding-based version each node serving as a grid gateway within the flooding region will help forwarding
geocast messages. In the ticket-based approach, geocast messages are still forwarded by gateway hosts, but
not all the gateways in the flooding region will do this job.

Trajectory-based forwarding (TBF)

Think in terms of an “agent”: Should travel around the network, e.g., collecting measurements.
• Random forwarding may take a long time
Idea: Provide the agent with a certain trajectory along which to travel
• Described, e.g., by a simple curve
• Forward to node closest to this trajectory

In the previous approaches, the destination region was – intuitively – conceived of as a more or less convex
region somewhere “far away”.
But this is not the only possible interpretation of geocasting as the somewhat different approach of TBF
shows.
Instead of trying to send a packet to some region far away, the region of interest can actually be a path in the
network.
This path, or trajectory of the packet, can be embedded into the packet as a parametric description of the
curve that the packet is supposed to follow; the parameter could be time or, preferably, the length of the path
that the packet has followed.
In this sense, trajectory-based routing combines aspects from source routing (as the trajectory is defined by
the source) and geocasting.

11. Explain the trajectory based forwarding and TORA unicast protocols of geographic routing.
Refer Q.10

12. With a neat diagram, explain the publish-subscribe interaction paradigm of data-centric WSN
networking.

The conceptual idea of the publish-subscribe interaction paradigm is essentially very simple: All nodes are
connected to a “software bus”. On this bus, data is made publicly available via a publish action; those nodes
that have previously announced their interest in that particular kind of data by an appropriate subscribe action
are then notified about the availability of this data. Figure 12.1 illustrates this concept; note how several
publishers can publish data of the same kind and how notifications can be delivered to various subscribers.
This concept of publication and subscription matches the requirement for a data-centric wireless sensor
network remarkably well. The publish/subscribe interaction pattern provides three essential properties
concerning the relationship between providers and subscribers of information:
Decoupling in space Publishers and subscribers need not to be aware of each other, they can be oblivious
of their mutual identities and numbers.
Decoupling in time Publishing and notification of data can happen at different times; the “software bus”
provides intermediate storage.
Decoupling in flows Interactions with the software bus can happen asynchronously without blocking.

13. What are the challenges of routing and forwarding techniques in WSNs? Discuss the importance of
multipath unicast routing.

14. Explain the Data-centric routing in wireless sensor network.

15. Explain the process and needs of the Data aggregation in wireless sensor network SPIN, ACQUIRE.