You are on page 1of 2

1

Homework 4
Communication Networks (EE 706), Spring’23

1) The following problem from Kurose and Ross, 6th edition, Chapter 4: P28, P29, P30, P31,
P32, P34 (on pp. 423-424) and P52 (on p. 428).
2) The following problems from Bertsekas and Gallager, Chapter 5 (pp. 479-481): 5.1, 5.2,
5.3, 5.4, 5.5, 5.7, 5.9 (first part), 5.10.
3) Give an example in which the asynchronous Bellman-Ford algorithm with poisoned reverse
is used and the count-to-infinity problem still occurs. Assume that all edge costs are positive.
4) (a) Let G = (N, E) be an undirected graph. Show that T is a minimum cost spanning tree
(MST) in G with edge costs {c(e) : e ∈ E} if and only if it is a MST in G with edge
costs {c(e) + k : e ∈ E}, where k is an arbitrary real number. Thus, an MST in a graph
with given edge costs remains an MST if the same constant k is added to every edge
cost.
(b) Show by example that a shortest path between two given nodes in a directed graph
with given edge costs does not necessarily remain a shortest path if the same constant
is added to every edge cost.
5) Consider the problem of finding shortest paths from all nodes to a given node u in a
weighted, directed graph G = (V, E). Suppose edges that are incoming into the destination
u (i.e., edges of the form (v, u)) may have negative cost, all other edge costs are non-
negative, and there are no negative-cost cycles. Consider the following statement: “Dijkstra’s
algorithm is guaranteed to correctly find shortest paths from all nodes to node u in this
graph.” State whether this statement is true or false and provide a proof (if true) or counter-
example (if false).
6) Consider a connected undirected graph, G = (N, E), where N is the set of nodes and E
is the set of edges, in which all edge costs are distinct (i.e, no two edge costs are equal).
Consider the following algorithm for finding a spanning tree. The edges in E are processed,
one at a time, in decreasing order of cost. For each edge, the edge is removed from the
2

graph if doing so does not disconnect the graph; otherwise the edge is retained in the graph.
This is continued until the remaining graph is a spanning tree. Show that at the end of the
algorithm, the remaining graph is a minimum spanning tree (MST). (Hint: Note that since
all edge costs are distinct, a unique MST, say T ∗ , exists. Show that the algorithm never
removes an edge from T ∗ .)
7) The distributed asynchronous Bellman-Ford algorithm executes in a network (directed graph)
starting from the initial conditions D(u) = 0 and D(v) = c(v, u) for each node v ̸= u,
where u is the destination node, D(v) is the distance estimate at node v and c(v, u) is the
cost of the edge from v to u if it exists and ∞ otherwise. Let D∗ (v) denote the shortest path
cost from v to u. Show that D(v) never decreases below D∗ (v) for any node v. Assume
that all edge costs are positive, and that if a link from node v to node w exists, then a link
from node w to node v also exists.

You might also like