1

Nash on the Crossroads of Information and

Game Theory
Anton Durbak
University of Illinois at Chicago
E-mail: [email protected]

Abstract
The purpose of this paper is to review elements of game theory as it applies to wireless
communication and interference channels. A special focus will be put on a series of John Nash’s
works [10] [11] [9], examining how his ideas, such as Nash’s Bargaining Solution and Nash’s
Equilibrium, are being used in a wide variety of channels. All of these elements will lead into a
survey of recent work that views concepts of information theory as part of a competitive game.

I. I NTRODUCTION
One can argue that in an increasingly growing world, it is imperative that we continually
strive to better the interactions that we encounter on a daily basis. When two or more
people find themselves in a conflict, certain strategies or rules are developed to find the best
possible solution. This is true in political standoffs, wartime negotiations, and even in our
technology. With every day, the multitude and power of our appliances and devices grow
as well. With the increase in number, comes an increase in interference from each other
and from other channels. Through the study of Information Theory, we begin to learn more
about the limits of different types of channels and different strategies that can be taken to
increase the capacity of a channel. These decisions can not always be made selfishly and
we must be able to program or control these links in a rational matter. Where information
theory leaves off, game theory becomes relevant.
Game theory is a field of applied mathematics that describes and analyzes interactive
decision situations. [12] Among the most important contributions to the field was that of
John Nash. He revolutionized the field by expanding the solutions necessary to generalizes
a strategy. Both Nash’s Bargaining Solution and Nash’s Equilibrium have been used as a
measure of multiple entity interactions [10] [11] [9].
We attempt to survey recent works and publications to see how game theory has been used
by different communication models, specifically wireless models and interference channels,
to expand the possibilities of the model and improve the now necessary focus of how these
channels interact with increasing number of channels fighting for a decreasing number of
resources.
Section II will give an overview of Game Theory, including a detailed look on Nash’s
Bargaining Solution (NBS) and Nash’s Equilibrium (NE).. Section III will survey recent
works who have used Game Theory in their analysis of Wireless models and Interference
channels.
 2

II. BASICS OF G AME T HEORY

To find the crossroads where game theory and information theory meet and how Nash’s
work is relevant, we need the underlying basics of game theory. Game theory is a field of
applied mathematics that describes and analyzes interactive decision situations. [12] While
the early work of game theory can be attributed to Cournot in 1838, the structure of the so
called normal form, that commonly characterizes this field, was done by John von Neumann
and Oskar Morgenstern nearly a century later.[8]
The normal form considered a model or situation as a game in which participating decision
makers are thought of as players. Each player has a strategy chosen from a set of available
actions to them. Each player can set a certain value to the outcomes of their actions which
has been described as a payoff or result of a utility function of this value.[14] Von Neumann
saw that these decisions needed to be strategically independent, until his and Morgenstern’s
analysis of a two person game. In their analysis the decisions would be done by coalitions,
rather than individuals, which came to be known as a cooperative game.

A. Nash Bargaining Solution

John Nash’s entrance into game theory began with proposing a solution to a cooperative
or bargaining game. Proposed in [9] and what would become known as Nash’s Bargaining
Solution, Nash attempted to find a solution to the general N-person cooperative game. He
saw that von Neumann’s analysis and solution only dealt with the two-person zero-sum
game. A zero-sum game is one where the total payoff between the two players would equal
zero. Neither player can increase or decrease the total value of the strategy.
Using von Neumann’s numerical utility representation, Nash would be able to produce a
solution through a graphical representation of the game by plotting each player’s utility gain.
Figure 1 is an example of this representation drawn by him in [9]. He made the assumption
that the solution set would need to be convex and compact, so that a point on a straight
line can always be obtained by a combination of two utility gains and so the set would be
bounded.
To produce a solution he made four more assumptions:
1) If there exists points α and β in the solution set where the utility function of β for all
players is greater than the utility function of α, then α could not be considered the
solution point. It is also known as Pareto optimality [5] and expresses the idea that
each individual wishes to maximize the utility to himself [9].
2) If the solution set is symmetric, then the solution point is on the line where U1 = U2 .
3) He maintained the linearity of the solution set, where if the same linear transformation
is performed then the solution is transformed appropriately [5].
4) If a larger set T contains a smaller solution set S and the solution point of T is in S,
then solution point T becomes solution point S as well.
Figure 2 was used by [9] to represent Nash’s assumptions.
Another concept that seems to be prevalent in many of the discussions of game theory is
the rationality of the players. Nash requires for the players to be rational, so as to be able
to accurately determine their objectives and goals in the bargain. Rationality also includes
that they are equal in bargaining skill and that they know fully the utility of each item to
each player[9].
 3

Fig. 1. Example of an application of Nash Bargaining Solution (NBS) given in [9]

B. Nash’s Equilibrium
In the formulation of his bargaining solution, Nash mentions ”an anticipation” that ”is
especially distinguished” [9], this was the anticipation of no cooperation between the players.
He would later formulate the idea in his doctoral thesis entitled ”Non-cooperative Games”.
The title would be used to describe a new type of game in which coalitions were not formed
and the players acted independently without collaborating communication [11]. The center
of this theory was that for a finite game, finite in time and strategies, there exists atleast
one equilibrium point. A Nash Equilbrium is a n-tuple point that corresponds to the mutual
best response for each player in terms of utility and one where a player cannot benefit
from changing strategies unilaterally[12]. In this equilibrium, while making an independent
decision, each player assumes that the other players will also pick a strategy that will induce
an Equilbria solution and has no incentive to stray from this choice.
To better present the concept of Nash’s Equilibrium and non-cooperative games, a model
was developed called the The Prisoner’s Dilemma. The version that will be used in this
survery was found in [14]. The story behind the game starts with the arrest and subsequent
interrogation of two prisoners in separate rooms. The district attorney offers a deal to each
prisoner if they confess to the crime and testify against the other. If they confess and the
other does not, then they would be set free as the other will serve up to 15 years in prison.
If they do not confess and the other does, then they will serve 15 years in prison. If neither
of them confess, then they will both serve one year in prison. If they both confess, then
they both serve 5 years in prison. Figure 3 shows this scenario.
 4

Fig. 2. Nash Bargaining Solution[9]

In this example, the choice of confession is considered the best response as your payoff can
not be worsened by the other’s decision. If each player plays the game with the knowledge
that the other will also play towards the mutual best response, then the tuple (Confess,
Confess) is considered the Nash Equilibrium. In a simple explanation, this holds true because
you cannot benefit from changing your strategy. As pointed out in [14], this strategy is
not considered Pareto optimal because from the prisoner’s point of view there were other
strategies that would yield less time in jail. With his equilibirum, Nash gave us a general
solution form to von Neumann’s model [8][7] and said that further work in game theory
would simplify to an analysis of equilibriums.

Fig. 3. Payoff Matrix of the Prisoner’s Dilemma[14]

III. F URTHER A PPLICATION OF NASH ’ S WORK

The further analysis of equilibriums can be applied beyond economic situations or math-
ematical proof. Nash’s Bargaining Solution (NBS) and Nash’s Equillibirum (NE) have been
 5

used in a wide variety of fields such as political science, sociology, biological science,
as well as more recently in the field of communication networks. This article will briefly
highlight the application of game theory in the realm of wireless models and interference
channels.

A. Wireless Models
The steady departure of wired appliances and the increasingly mobile world has made the
study of wireless technology very important. As this mobile world expands, the resources
become more limited and thus creates a competitive environment to share these properties
such as power, bandwidth, etc. Protocols and strategies are formulated as the interaction two
wireless entities is seen in the sense of a game. Each party or node enters into a situation in
which recent work shows NE or NBS is the best strategy or solution. In [15], a cooperation
strategy is developed in relay networks based on the Nash Bargaining Solution, where the
utility function is used to find a cooperative strategy to allocate bandwidth. The solution
solves the question of when and how to cooperate.
[13] takes a similar approach in their analysis on spectrum sharing networks, as they find
NBS to be a compromise of fairness and efficiency. The game is modeled using the Shannon
capacity
K
BX H k pk
Ci (p) = log2 (1 + σ2 Pii i k k )
K k=1 K
+ j6i Hji pj
where pki is the power transmitted by node i and Hjik as the channel gain from j to the
receiver of i, as the utility function.
[14] designed a NE backoff strategy for Wireless ad Hoc Networks. Wireless Ad Hoc
Networks are a collection wireless nodes that have no central infrastructure and require a
protocol to assist in the passing of information though the network. They formulated the
problem similarly to the Prisoner’s Dilemma, where the node have a choice of transmit or
not transmit and each pairing has a certain payoff and/or penalty, such as a utility function
of a failed transmission or delay sensitivity. They applied a Nash strategy to a TCP session
and found a fair backoff strategy between nodes. A thorough breakdown of a wireless ad
hoc network and how they can be modeled as games by OSI layer, is featured in [12].

B. Interference Channels
The work on wireless models attempt to formulate strategies to combat problems and
conflicts with limited resources. This conflict allows us to view the nodes as players in
a game who are trying to find their optimal solutions or capacities. Finding the capacity
region of Interference channels is still a problem in information theory. There has been some
success in achievable regions such as the work of Han and Kobayashi, who found a new
achievable rate interference channels with ”less” strong interference using a polymatroidal
structure [4]. While their work is constantly being updated and improved upon [3], new
research has been looking at the problem through a game theoretic viewpoint. The Han-
Kobayashi uses a scheme that requires that the two channels split their information into a
common and private stream. The common stream is encoded to lessen the interference seen
by the other user. The choice of how to encode or decode along with the desire to maximize
one’s rate is a conflict that can be seen as a non-cooperative game [2].
 6

If capacity regions can be viewed as a Nash solution region with an equilibrium point,
then there must be a pairing of endcoding strategies, for instance, that either channel would
not want to deviate from. Figure 4 shows that if we characterize α as an interference level,
or ratio of Signal to Noise Ratio (SNR) and Interference to Noise Ratio (INR), then as α
increases the area seen as the intersection of the deterministic channel capacity region C and
NE bounds ”box”B also increases. The area always has a point on the sum-rate boundary
of C [2].

Fig. 4. Examples of CN E for a symmetric deterministic channel with normalized cross gain α [12]

The characterization of interference channels as competitive games is not new, but recently
has been gaining momentum. It has most likely done so because of the interest on the
interaction of individual links or channels and the growth of non-cooperative solutions, such
as iterative water-filling, as a viable solution for multiple access channels and broadcast
channels [5][6]. [5] shows the development of a game theoretic region of achievability,
because information theory will give you a region that allows all points on the boundary on
the channel may be achievable, but not realistic as a link may choose to deviate from the
encoding scheme. Using the concepts of NE and NBS, one would be finding the rate pairs
least likely to be deviated from.
Not every application of game theory is automatically linked with Nash’s contributions
and proved successful. [1] found multiple Nash Equilibirum points in different evaluation
regions when trying to formulate a specturm sharing scheme of two receivers and two
carriers. They found that a scheme following a Stackelberg game model, where a certain
operator is chosen as a leader, produced better results and more achievable. Figure 5 shows
their simulation results.

IV. C ONCLUSION
It has increasingly become apparent that game theory can be applied to many different
fields. Most recently, there has been a growing movement to use elements of game theory,
 7

Fig. 5. [1]

especially the various contributions of John Nash, to combat some of the limitations of
Information theory. What NE and NBS provide, is as stable-set of solutions that have
withstood decades of advancements to stand relevant. This new insight can only expand
our view to see channels and links as players in a game and that John Nash might have
already solved it.

R EFERENCES
[1] M. Bennis, M. Le Treust, S. Lasaulce, M. Debbah, and J. Lilleberg, “Spectrum sharing games on the interference
channel,” in Game Theory for Networks, 2009. GameNets ’09. International Conference on, May 2009, pp. 515–522.
[2] R. Berry and D. Tse, “Information theory meets game theory on the interference channel,” in Proceedings of the
IEEE Information Theory Workshop (ITW), June 10-12 2009.
[3] Y. Chao, B. Chen, and J. Zhang, “A new achievable rate region for the interference channel with common information,”
in Wireless Communications and Networking Conference 2007, March 11 - 15 2007, pp. 2071–2075.
[4] T. S. Han and K. Kobayashi, “A new achievable rate region for the interference channel,” IEEE Transactions on
Information Theory, vol. 27, no. 1, pp. 49–60, January 1981.
[5] A. Leshem and E. Zehavi, “Bargaining over the interference channel,” in Information Theory, 2006 IEEE International
Symposium on, July 2006, pp. 2225–2229.
[6] ——, “Game theory and the frequency selective interference channel,” Signal Processing Magazine, IEEE, vol. 26,
no. 5, pp. 28–40, September 2009.
[7] G. J. Mailath, “Do people play nash equilibrium? lessons from evolutionary game theory,” Journal of Economic
Literature, vol. 36, no. 3, pp. 1347–1374, September 1998.
[8] R. B. Myerson, “Nash equilibrium and the history of economic theory,” Journal of Economic Literature, vol. 37,
no. 3, pp. 1067–1082, September 1999.
[9] J. Nash, “The bargaining problem,” Econometrica, vol. 18, pp. 155–162, April 1950.
[10] ——, “Equilibrium points in n-person games,” Proceedings of the National Academy of Sciences, vol. 36, no. 1, pp.
48–49, 1950.
[11] ——, “Non-cooperative games,” The Annals of Mathematics, vol. 54, no. 2, pp. 286–295, 1951.
[12] V. Srivastava, J. Neel, A. Mackenzie, R. Menon, L. Dasilva, J. Hicks, J. Reed, and R. P. Gilles, “Using game theory
to analyze wireless ad hoc networks,” IEEE Communications Suverys and Tutorials, vol. 7, no. 4, pp. 46–56, 2005.
[13] J. Suris, L. DaSilva, Z. Han, and A. MacKenzie, “Cooperative game theory for distributed spectrum sharing,” in
Communications, 2007. ICC ’07. IEEE International Conference on, June 2007, pp. 5282–5287.
[14] Y. Xiao, X. Shan, and Y. Ren, “Game theory models for ieee 802.11 dcf in wireless ad hoc networks,” IEEE
Communications Magazine, vol. 45, no. 3, pp. S22–S26, March 2005.
[15] Z. Zhang, J. Shi, H.-H. Chen, M. Guizani, and P. Qiu, “A cooperation strategy based on nash bargaining solution in
cooperative relay networks,” IEEE Transactions on Vehicular Technology, vol. 57, no. 4, pp. 2570–2577, July 2008.